Photoemission From A Two Electron Quantum Dot Subinoy Das Department of Physics, Indian Institute of Technology, Kanpur, U.P.–208016, India 2 0 email: [email protected] 0 2 Pallab Goswami n Department of Physics, Indian Institute of Technology, Kanpur, U.P.–208016, India a J email: [email protected] 8 1 and ] J.K. Bhattacharjee1 l l Department of Theoretical Physics, Indian Association for Cultivation of Science a h Calcutta – 700032, India - s email: [email protected] e m Dated:14th January, 2002 . t a m - d We consider photoemision from a two-electron quantum dot and find analytic expression for n the cross-section. We show that the emission cross-section from the ground state as a function of o c the magnetic field has sharp discontinuities corresponding to the singlet-triplet transitions for low [ magnetic fields and the transitions between magic numbers (2J+1) for high magnetic field. We also 1 v findthecorrections tothephotoemissioncross-section fromamorerelisticquantumdothavingafinite 1 4 thickness due to nonvanishing extent of electron wave function in the direction perpendicular to the 3 1 plane in which 2-D dot electrons are usually confined. 0 2 0 Artificial atoms or quantum dots, which are essentially electrons confined in a two-dimensional / t a region with a magnetic field in the third direction have been the subject of intense experimental and m theoretical activity over thelast few years [1]. Theartificial ”hydrogen atom” (single electron confined - d in a circular region by a harmonic potential with a magnetic field in the perpendicular direction) n o was solved for the eigenvalues and eigenfunctons over seventy years ago by Fock [2]. The levels were c : experimentally observed [3] morethan fifty years later with the advent of quantum dots. Theartificial v Xi ”heliumatom”(twoelectronsconfinedinacircularregionbyaharmonicpotentialwithamagneticfield r in the third direction) was tackled sixty years after the hydrogen atom. Maksym and Chakraborty a [4] and Wagner, Merkt and Chaplik [5] worked out the energy levels and found an incredibly rich structure. The ground state can change its parity as the magnetic field is changed and there can be singlet-triplet transition. However, spectroscopic studies did not reveal the spectacular features because of an effect noted by these authors [4], [5]. Instead, they relied on certain thermodynamic 1Authorto whom all correspondence should beaddressed 1 measurements to support the energy level structure that they obtained. This ”helium atom” problem was exactly solved a few years later by Dineykhan and Nazmitdinov [6], giving answers which agreed with the findings of Maksym and Cahkraborty [4] and Wagner et al [5]. In this note, we wish to point out that photoelectric efect is an experiment that would probe the different transitions in the ground state energy as the magnetic field is varied. We will show discontinuities in the cross-section as a function of the magnetic field corresponding to the singlet-triplet transitions and find the corrections due to finite thickness by imposing a stronger harmonic confinement in the z-direction. Considering two electrons in a circular dot with a magnetic field perpendicular to the circular region, we can write the hamiltonian, 2 ¯h2 ω 1 e2 1 H = [ 2+ c( i¯h )+ m ω2ρ2]+ (1) −2m ∇j 2 − ∇φj 2 ∗ c j 4πǫǫ ρ~ ρ~ jX=1 ∗ 0| 1− 2 | where ρ~ and ρ~ are the two dimensional position vectors of the two electrons, ω is the cyclotron 1 2 c frequency, m is the effective mass of the electron in the semiconductor and Ω2 = (ω2 + ωc2). Using ∗ 0 4 the COM coordinate ρ~ = 1(ρ~ +ρ~ ) and the relative coordinate ρ~ = (ρ~ ρ~ ), the hamiltonian c 2 1 2 rel 1 − 2 clearly splits into H = H +H , where H dependsonly on the center of mass coordinates. This part c rel c of the hamiltonian is a purely single electron hamiltonian and can be exactly solved. The part H rel on the other hand involves the important Coulomb repulsion and is responsible for the rich structure of the energy spectrum. Now if there is an external electric field of long wavelength (far infrared spectroscopy) imposed on the system, then the dot size being much smaller than the wavelength, there is no appreciable change in the elctric field across the sample and hence the contribution to the hamitonian is eE~ (ρ~ + ρ~ )exp(iωt) = 2eE~ ρ~ exp(iωt) where E~ is a constant electric field. 1 2 c · · Thus, this perturbation does not couple to the ρ~ -dependent part and is completely blind to the rich rel structure coming from. Now, if photoelectric effect is what we are interested in, then the external electromagnetic field can be considered as coming from a vector potential A~ext = nˆA exp(iωt) and 0 the perturbation hamiltonian H˜ is p~A~ext to the lowest order, which can be written as ·m∗ A H˜ = 0[p~ nˆexp(i~k ρ~ )+p~ nˆexp(i~k ρ~ )]exp( iωt) (2) 1 1 2 2 m · · · · − ∗ and since the dipole approximation is no longer required (i.e. long wavelength condition is not im- posed), we have the contribution to the ionization cross-section coming from both H and H . ρ~c ρr~el The photoemission cross-section involves the matrix element ψ (ρ ,ρ )H˜ψ (ρ ,ρ )d2ρ d2ρ . f∗ 1 2 i 1 2 1 2 The initial state is the ground state of the dot and can be writtenR as φ1(ρ~c)ψ2(ρ~rel) while the fi- nal state corresponds to a free electron of wave number ~q and a bound electron in the lowest energy state of a single electron quantum dot i.e.ψ (ρ~ ,ρ~ ) = φ(ρ~ )exp(iq~ ρ~ ). It is this integral which is f 1 2 1 2 · sensitive to the nature of initial state. As the initial state undergoes a singlet-triplet transition, the orbital parity of the spatial wave function changes and there is a sudden jump in the matrix element 2 and hence the cross-section as a function of the magnetic field will show jumps at the singlet-triplet transitions. To establish the above result, the most important fact that we need to know is the two-electron wave function for the ground state. We have found an extremely accurate variational wave function for the ground state [7]. This wave function (normalized) is 1 ρ ρ2 ρ2 Ψ(ρ~ ,ρ~ ) = ( rel)l exp( c )exp( rel)exp( ilφ ) (3) c rel 2(|2l|+1)πa˜Hβ(Γ( l +1))12 β || −4a˜H2 −4β2 − rel | | where a˜ 2 = ¯h and β is a variational parameter fixed by the energy minimization condition H 4m∗Ω a˜ Γ( l +1) x4 √2 H | | 2 x 1 = 0 (4) − a Γ( l +2) − ∗ | | where x = β and a = 4πǫǫ0¯h2 is the Bohr radius of the semicoductor material (energy spectrum is 2a˜H ∗ m∗e2 plotted in fig.1). The final state corresponds to one free electron, and one electron in the ground state of single electron quantum dot. The matrix element for photoemission can now be written as e 2π¯h < f H˜ i >= [ ψ⋆ (ρ~ ,ρ~ )exp(i~k ρ~ )( i¯hnˆ )ψ (ρ~ ,ρ~ )dρ~ dρ~ +(1 2)] (5) | | m s ω final 1 2 · 1 − ·∇1 initial 1 2 1 2 ←→ ∗ ZZ The two contributions to < f H˜ i > will be equal and hence we need to evaluate only one integral I | | which is e 2π¯h (i)2¯hnˆ (~k ~q) ρ2 ρ2 I = · − exp[i(~k ~q) (ρ~ )]exp[ 2 ]exp[ c ] m∗s ω 2(|l|+3)π3a2Ha˜H2β2Γ(| l | +1)AZZ − · 1 −4a2H −4a˜H2 × ρ2 q ρ exp[ rel ilφ ]( rel)ld2ρ d2ρ (6) −4β2 − rel β || 1 2 where a2 = 2a˜ 2. Now, H H ρ2 ρ2 ρ 1 1 1 1 exp[ c ]exp[ rel ilφ ]( rel)l = exp[ (ρ2+ρ2)( + )]exp[ ρ ρ ( ) −4a˜ 2 −4β2 − rel β || − 1 2 16a˜ 2 4β2 − 1 2 8a˜ 2 − 2β2 × H H H ρ exp[ iφ ] ρ exp[ iφ ] cos(φ1 φ2)]( 1 − 1 − 2 − 2 )(|l|) (7) − β After putting the expression in equation (7) in equation (6) and writing ~k ~q = K~, we note that − K~ ρ~ = Kρ cos(φ η). Now the angular integrations in the above integral can be performed by 1 1 1 · − using, as necessary,the following identities: exp[ibcos(φ φ )] = ∞ (i)mJ (b)exp[im(φ φ )] 1 2 m 1 2 − − m= X−∞ exp[ ccos(φ φ )] = ∞ (i)2mI (c)exp[im(φ φ )] 1 2 m 1 2 − − − m= X−∞ 2π dφ exp[i(m n)φ] = δ (8) m,n 2π − Z0 3 After carrying out the angular integrals with the help of binomial expansion (ρ1exp[−iφ1]β−ρ2exp[−iφ2)|l| = |l| Γ(t+Γ1)(Γ|l(|l+1)t+1)ρ|1lβ|−ltρt2 exp[−i(| l | −t)φ1]exp[−itφ2] (9) Xt=0 | |− || and writing the constant term outside as C we are left with radial integral parts l C l < f | H˜ |i > = 2βll ρ1dρ1ρ2dρ2 Ct|l|ρ|1l|−tρt2exp[−ρ21b]exp[−ρ22c](−1)t(i)|l|× || Z t=0 X exp[ i l η(2π)2J (Kρ )I (dρ ρ )] (10) l 1 t 1 2 − | | = (2π)2Cl l ( 1)tCl(i)lexp[ ilη] 1 ∞ ∞ ( 1)p Γ(p+q+ |l | +1) 4βlbc − t − (√b)l t√ct − Γ(p+1)Γ(q+1)Γ( l +p+1) × || Xt=0 − pX=0qX=0 | | K d ( )2p+l( )2q+t (11) || 2√b 2√bc After some manipulations and using the generating function of associated Laguerre polynomial we find the closed form (2π)2C K lπ (1 d)l K2 < f H˜ i>= l( )l exp[ ilη+i ] − 2c || exp[ ] (12) | | 4βlbc 2b || − 2 (1 d2 )l+1 −4b(1 d2 ) || − 4bc || − 4bc 2 2 2 where b = 1+x2, c = 3+x2 and d = 1−x2. From this we get the expression for the differential 8a2 8a2 4a2 H H H cross-section dσ dφ~q = 28+3|l| [1 ( l +1)Ω+l ωc]sin2θcos2φ(x2+2)2|l|(KaH)2|l| exp[ 2K2a2H(3x2+2)] (13) σ Γ( l +1) − | | ω 2 ω x2l 2(x2+6)2l+2 − (6+x2) 0 | | ||− || Here ~q is the wave vector of the emitted electron, ~k is the wave vector of the incident photon, h¯K~ = ¯h(~k ~q) is the momentum transferred and θ and φ are respectively the angles ~q makes with~k and~knˆ − plane where nˆ is the unit polarization vector of the incident photon. σ = e2 (me)2(2π) is a constant 0 ca∗2 m∗ ¯h extracted to express the differntial cross-section expression in a dimensionless form. So, K2 = k2+q2 2kqcosθ (14) − holds. and cosφ = sinθ cos(φ φ ) and cosθ = sinθ cos(φ φ ). Therefore, it is very clear nˆ nˆ − q~ ~k ~k − q~ from the expression for the differential cross-section that it depends significantly on the direction of incidence and polarization. This for some simple cases can be illustrated easily. If ~k is parallel to z-axis then cosθ = 0 and cosφ = cosφ or cosφ = sinφ as nˆ is parallel to x or y-axis. So, the q~ q~ angular distribution is proportional to cos2φ or sin2φ and if it is the case of circular polarization q~ q~ then the angular distribution is proportionalto (cos2φnˆcos2φq~+sin2φnˆsin2φq~) and only if nˆ = (xˆ√±2iyˆ) then it becomes isotropic. But, when the ~k lies in the x-y plane then with all the cases of circular polarization we shall have angular dependence. It also becomes apparent from the expression that emission count is larger in the direction of polarization compared to other cases and if the photon 4 is linearly polarized in the z-direction then there is no emission. So, depending on the ’l’ values of the ground state as a function of magnetic field the cross-section would have different angular distribution as well as discontinuities characterizing transitions of the ground state. In this case it will also be found that if Based on these one can probe now these transitions experimentally. For this purpose one has to choose carefully the magnetic field strength, incident photon frequency and the abovementioned directions. We plot the dimensionless expression of the differential cross-section (fig.2) for the transitions l = 0 l = 1 and l = 1 l = 2 which take place (for our chosen system size → → ¯hω = 4meV)at1.3T and6.1Trespectively when~k zˆand~q nˆ (i.e. whenitismaximum). Fromthe 0 k k plot it can be seen that there are discontinuities at the magnetic field strengths where singlet-triplet transitions are taking place and where for the cases of l = 0 and l = 1 the cross-section increases with magnetic field strength for l = 2 it decreases. This can be understood on the physical ground in analogy with atomic photo-effect. As magnetic field strength increases wave function is compressed more and more that means electrons become more tightly bound and emission increases. When transition takes place depending on the energetics electron wave function becomes further compressed which manifests in the x-values for different ’l’ values and the emission count shows a jump. But, after sufficient increase in the magnetic field strength the energy value also changes considerably and with it ionization energy changes. Thus upon keeping the incident frequency fixed after a certain range of magnetic field the cross-section decreases with increasing field strength. In the real dots there is finite extent of the wave function in the z-direction and for the experi- ments it also needs to be considered. We teke this into account by considering a stronger harmonic confinement in the z-direction. with this the hamiltonian gets modified with the term ¯h2 d2 1 H = + m ω2z2 (15) z −2m dz2 2 ∗ z ∗ and the Coulomb term becomes e2 1 . With the H part we have the wavefunction modified by 4πǫǫ0 r~1 r~2 z | − | 1 z2 z ψ (z) = exp( )H ( ) (16) 3 2nΓ(n +1)π1/2λ −2λ2 n λ z q and the energy is modified by the term E = (n + 1)h¯ω . When solved variationally (introducing z z 2 z two parameters β and β instead of two independent oscillator lengths) we have for the two-electron 1 2 dot the following coupled characteristic equations 2 a a Γ( l +1) 3 5 x a x4[1 ( H)( H )3 | | 2F ( ,2+ l , + l ,1 2( )2( H )2)] = 1 (17) − y3 a λ Γ(5+ l ) 1 2 | | 2 | | − y λ ∗ rel 2 | | rel λ Γ( l +1) 1 3 x a x2 a2 Γ( l +2) y4 y( rel) | | 2F ( ,1+ l , + l ,1 2( )2( H )2)+2 H | | − a Γ(3+ l ) 1 2 | | 2 | | − y λ y a λ Γ(5+ l ) × ∗ 2 | | rel ∗ rel 2 | | 3 5 x a 2F ( ,2+ l , + l ,1 2( )2( H )2) = 1 (18) 1 2 | | 2 | | − y λ rel 5 where oscilator lengths are related as λ2 = λ2rel = ¯h and y = β2 . When solved numerically, 2 m∗ωz λrel from the energy spectrum it is found that transitions of the ground state are taking place at higher magnetic field values (l = 0 l = 1 and l = 1 l = 2 occur at 3.1T and 11.2T respectively for our → → chosen ratio ωz = 9) as it becomes evident from fig.5 and the dimensionless differential cross-section ω0 now becomes ddΩσ~q = 28+3|l| [1 ( l +1)Ω + l ωc]sin2θcos2φ(x2+2)2|l|(KρaH)2|l| exp[ 2Kρ2a2H(3x2+2)] σ Γ( l +1) − | | ω 2 ω x2l 2(x2+6)2l+2 − (6+x2) × 0 | | ||− || 8 1(λrel) 1 exp[ K2λ2 ( 1 + (2− y12)2 )] (19) √2πy a ( 3 + 20 +8) − z rel (3+ 1 ) 4(3+ 1 )(1+ 1 ) (2 1 )2 ∗ y4 y2 y2 y2 y2 − − y2 Here, K2 = K2+K2 and K2 = K2 K2 = K2cos2θ and ρ x y z − ρ K~ kcosθ qcosθ cosθ = ~k − q~ K~ k2+q2 2kqcosθ − cosθ = sinθ sinθ cos(φ φ )+cosθ cosθ p ~k q~ ~k − q~ ~k q~ sinθcosφ = sinθ sinθ cos(φ φ )+cosθ cosθ (20) nˆ q~ nˆ q~ nˆ q~ − Putting these in the expression for differential cross-section and by explicit integration one can find the total cross-section. For the incidence direction parallel to z-axis and emission in the direction of polarization we show the plot for this modified differential cross-section (fig.7) with the earlier photon energy. From the plot it is seen that the cross-section value is now sufficiently suppressed compared to earlier case as energy of the corresponding states have significantly. Also, due to this reason where in the earlier 2-D case decrease in the cross-section as a function of field strength started for l = 2 for the 3-D dot it started decreasing right from the l = 1 state after certain amount of increase and at l = 1 l = 2 transition the count decreased in contrast to increase in 2-D situation. → We conclude by considering the feasibility of an experiment which would detect the above effect. Thefirstthingtonoteisthatwewantacouplingofrelativeco-ordinatetotheexternalelectromagnetic field.Consequently the wavelength of the radiation must be smaller than the size of the dot which can be of the order of 100 nanometers. This implies that we will be dealing with energetic photons and photo-emission will be from the bound state in the dot into vacuum. There will be emission from the semiconductor as well and we would have to substract this background contribution.This can be achieved by studying the angular distribution.The distribution of the electron knocked out from the rest of the solid will be isotropic where as the distribution of the electrons coming from the dot would have a zero in the forward direction. This enables one to know the background which can now be substracted to find the cross-section of the photoemission from the dot. Another point is that as the magnetic field applied to the dot can be considered to be local compared to the bulk, with the change of magnetic field while there will be change in the angular dependence and counts from the dot, that from the bulk would remain unchanged. 6 An alternative technique is to do the experiment first with the GaAs layer without dot. This determine the background. Next,one can repeat it with single-electron dots and finally with two- electron dots. The difference between the two-electron dot and the single-electron dot will exhibit the correlation effect. From the plot for angular distribution for differential cross-section (fig.3,fig.4) it becomes clear thatwhentransitionsoccurtheangulardistributionalsochangesandmaximumofemissionisobtained for a definiteangle of incidence and this angle changes with transitions. So, at the time of experiments keeping the photon enegy fixed and varying the angle of incidence and observing the change in angle correspondingtomaximaduetochangein’l’states,thetransitionscanbeprobed. Themostsignificant point which should be stressed is that for certain angle of incidence (as evident from fig.3 and fig.4) the count does not change even at the points of transitions. So to detect the transitions by the discontinuities in the cross-section, this angle should be carefully avoided. We have already argued above that if photon energy is kept fixed then all the transitions can not be observed properly. So, depending on the system size and field strength and ’l’ values of the states angle of incidence and photon energy have to be carefully chosen and maximum of count will always be obtained in the direction of polarization. Is the effect big enough to be measured? The scale for the cross-section of emission from the dot when compared to the scale for the emission from the bound state of an atom is the ratio m a˜H2 m∗ β2 which is much greater than unity and from our analytic expression for the emission cross-section it is evident that the ratio σ is a large number. So the scale for the emission cross-section is bigger when 0 emission from the dot is involved. This should make the experiment quite feasible. References [1] T.Chakraborty, ”Quantum Dots” Elsevier Publications, North Holland (1999) [2] V.Fock, Z.Phys., 47,446 (1928) [3] R.C.Ashoori, H.L.Stormer, J.S.Weiner, L.N.Pfeiffer, K.W.Baldwin and K.W.West, Phys. Rev. Lett. 71, 613 (1993) [4] P.A.Maksym and T. Chakraborty, Phys. Rev. Lett.65 108 (1990), Phys. Rev. B45, 1947 (1992) [5] M.Wagner, U.Merkt and A.V.Chaplik, Phys Rev. B45 1991 (1992) [6] M.Dineykhan and R.G.Nazmitdinov, Phys.Rev. B55 13707 (1997) [7] P.Goswami, S.Das and J.K. Bhattacharjee (cond-mat/0201225) 7 6.0 5.5 5.0 l=0 ) Z ( 4.5 l=3 y g r e n 4.0 e 3.5 l=2 3.0 l=1 2.5 0 2 4 6 8 10 12 B(T) Figure 1: Z = Erel+Espin vs. B(T) is plotted for different ’l’ values and l = 0 l = 1, l = 1 l = 2 ¯hω0 → → transitions are taking place at B = 1.3T and B = 6.1T respectively and also other energy level crossings are present. 8 2 n o i t c 1.5 e s - s s o r c 1 l a i t n e r 0.5 e f f i D 0 0 2 4 6 8 Magnetic field B dσ Figure 2: dφ~q vs. B(T) is plotted for different ’l’ values of the ground state as the B is varied and σ0 showing discontinuities as characteristic of the transitions. From the plot it is found that percentage change for l = 1 l = 2 is 4.5 times smaller compared to l = 0 l = 1 transition and also it → ≈ → is evident from the plot that at a fixed frequency behaviors of different ’l’ cross-sections are going to change and for this reason both ω and B have to be varied to observe all the transitions properly. 9 e c 1.5 n e 1.25 d n e 1 p e D 0.75 r a 0.5 l u g 0.25 n A 0 0 0.25 0.5 0.75 1 1.25 1.5 theta dσ Figure 3: Difference in angular dependence of dφ~q for transition l = 0 l = 1 is shown σ0 → 10