MIT Department of Chemistry 5.74, Spring 2004: Introductory Quantum Mechanics II Instructor: Prof. Andrei Tokmakoff p. 28 THE RELATIONSHIP BETWEEN U(t,t0) AND cn(t) For a time-dependent Hamiltonian, we can often partition H = H0 + V (t ) H0 : time-independent; V (t ) : time-dependent potential. We know the eigenkets and eigenvalues of H0 : H0 n = En n We describe the initial state of the system (t = t 0 ) as a superposition of these eigenstates: ψ ( t 0 ) = ∑ cn n n For longer times t , we would like to describe the evolution of ψ in terms of an expansion in these kets: ψ (t ) = ∑ cn (t ) n n The expansion coefficients ck (t ) are given by ck (t ) = k ψ (t ) = k U (t,t 0 )ψ (t0 ) Alternatively we can express the expansion coefficients in terms of the interaction picture wavefunctions bk (t ) = k ψ I (t ) (This notation follows Cohen-Tannoudji.) Notice ck ( t ) = k ψ ( t ) = k U 0 U I ψ ( t 0 ) = e−iωk t k U I ψ ( t 0 ) = e−iωk t b k ( t ) 2 2 so that bk (t ) = ck (t ) . Also, bk (0 ) = ck (0) . It is easy to calculate bk (t ) and then add in the extra oscillatory term at the end. p. 29 Now, starting with ∂ψ I i ∂t = VI ψ I we can derive an equation of motion for bk ∂ bk = k VI U I ψ I ( t 0 ) ∂t i inserting ∑n ψ I (t0 ) = ∑ bn n n = ∑ k VI n n U I ψ I ( t 0 ) n =1 n n = ∑ k VI n b n ( t ) n i ∂b k = ∑ Vkn ( t ) e− iωn k t b n ( t ) ∂t n This equation is an exact solution. It is a set of coupled differential equations that describe how probability amplitude moves through eigenstates due to a time-dependent potential. Except in simple cases, these equations can’t be solved analytically, but it’s often straightforward to integrate numerically. Exact Solution: Resonant Driving of Two-level System Let’s describe what happens when you drive a two-level system with an oscillating potential. V ( t ) = V cos ωt = Vf ( t ) This is what you expect for an electromagnetic field interacting with charged particles: dipole transitions. The electric field is E ( t ) = E 0 cos ωt For a particle with charge q in a field E , the force on the particle is F = qE which is the gradient of the potential p. 30 Fx = − ∂V = qE x ∂x ⇒ V = −qE x x qx is just the x component of the dipole moment µ . So matrix elements in V look like: k | V(t) | = −qE x k | x | cos ωt More generally, V = −E ⋅ µ . So, V ( t ) = V cos ωt = − E 0 ⋅µ cos ωt . Vk ( t ) = Vk cos ωt = − E 0 ⋅µ k cos ωt We will now couple our two states k + starts in with the oscillating field. Let’s ask if the system what is the probability of finding it in k at time t ? The system of differential equations that describe this situation are: i ∂ b k ( t ) = ∑ b n ( t ) Vkn ( t ) e−ωnk t ∂t n = ∑ b n ( t ) Vkn e − iωnk t × 12 ( e− iωt + e+ iωt ) n i ω −ω t i ω +ω t i b k = 12 b Vk e ( k ) + e ( k ) + 12 b k Vkk eiωt + e− iωt = (1) and (2) i ω −ω t i ω +ω t i b = 12 b V eiωt + e− iωt + 12 b k V k e ( k ) + e ( k ) or = (3) and (4) e − i( ωk +ω) t + e− i( ωk −ω) t We can drop (2) and (3). For our case, Vii = 0 . We also make the secular approximation (rotating wave approximation) in which the i ω k + ω )t nonresonant terms are dropped. When ω k ≈ ω , terms like e ±i ωt or e ( rapidly and so don’t contribute much to change of cn . oscillate very p. 31 So we have: bk = −i 2 b Vk e ( i ωk −ω) t b = −i 2 bk V k e (1) − i ( ωk −ω) t (2) Note that the coefficients are oscillating out of phase with one another. Now if we differentiate (1): b Vk ei( ωk −ω) t + i ( ωk − ω) b Vk ei( ωk −ω) t bk = −i 2 b = 2i − i ω −ω t bk e ( k ) Vk (3) Rewrite (1): (4) and substitute (4) and (2) into (3), we get linear second order equation for b k . 2 V b k − i ( ωk − ω) b k + k 2 b k = 0 4 This is just the second order differential equation for a damped harmonic oscillator: ax + bx + cx = 0 x=e − ( b 2a ) t ( A cos µt + B sin µt ) µ= 4ac − b 2 1 2 (remember b k (0)=0 and b (0)=1) With a little more work, we find Pk = b k ( t ) = Vk 2 ΩR = P = 1 − bk 1 2a 2 1 2 Vk 2 Vk 2 2 sin 2 Ω r t + 2 ( ωk − ω) + 2 ( ωk 2 − ω) 2 1 2 p. 32 Amplitude oscillates back and forth between the two states at a frequency dictated by the coupling. Resonance: To get transfer of probability amplitude you need the driving field to be at the same frequency as the energy splitting. Note a result we will return to later: Electric fields couple states, creating coherences! On resonance, you always drive probability amplitude entirely from one state to another. 1 Pkl on resonance ωkl − ω = Vkl 0.5 0 0 large detuning Efficiency of driving between t π / Vkl and k states drops off with detuning. Pmax 2 Vkl / ω kl