betaFeedback

WorksPhysics and Astronomy

Quantum Thermal Logic Gates

Abstract

We propose a new concept for quantum thermal logic gates – analogous to classical electronic logic gates – that exploit the heat current in a coupled quantum-dot system tunnel-coupled to metallic thermal reservoirs for logic operations in quantum circuits. We obtained a remarkable one-to-one correspondence with the structure of classical electronic logic gate circuits. An experimental setup is presented that demonstrates a realizable nano-electronic quantum circuit architecture for implementing such quantum thermal logic operations.

Introduction

In the emerging era of quantum computation, achieving error-free and thermally efficient quantum circuits is the ultimate goal [1, 2, 3, 4, 5, 6]. Toward this effort, recent advancements in building compact quantum circuit architecture with unprecedented control over individual quantum systems are remarkable [7, 8, 9, 10, 11, 12]. These include the development of quantum-advantageous devices on multiple platforms, e.g., nano-electronic quantum transistors [13, 14, 15, 16, 17], diodes [18, 19, 20, 21], capacitors [22], batteries [23, 24], heat engines and refrigerators [25, 26, 27, 28], transformers [29], etc. In addition to developing innovative and compact designs of nano-electronic quantum circuits, managing thermal dissipation during their operation is another important issue [30] for ensuring reliable and error-free device performance. To address these challenges, based on local thermal-measurement techniques [31, 32, 33, 34, 35, 36], various quantum devices with optimized thermal operations have been proposed and demonstrated [37, 38]. Some noteworthy developments include strong electron interactions mediated enhanced heat-flow in single-electron junctions [39, 40, 41], near-optimal quantum heat engines with efficiencies close to the Carnot limit [42, 43], and quantum heat valves that regulate heat-flow using quantum levels [44, 45, 46], among others. These advances in controlling heat-flow and dissipation in quantum devices have already set the stage for further quantum thermodynamic explorations toward integration in quantum circuits [47, 48]. However, establishing a quantum analog of classical electronic logic-gates for heat-based logic operations, while being a crucial element in integrated quantum circuits, remains largely unexplored.

In this letter, we introduce for the first time the concept of quantum thermal logic gates (QTLG), which exploit temperature gradients and heat currents in a nano-electronic coupled quantum-dot (CQD) [19, 49, 50, 51, 52] junction to perform logical operations — analogous to how electronic logic gates process electrical signals. Such devices not only promise new modes of energy-efficient computation but also hold potential for intelligent thermal management in nano-electronic architectures. With the growing demand for sustainable energy solutions and waste-heat recovery in nanoscale systems, we focus on controlling thermal transport at the quantum level by exploring the fundamental principles and minimal models required to realize QTLG — providing a step toward programmable, experimentally realizable heat-based computation and energy-efficient quantum technologies [53, 54].

General Model

The general model of the proposed QTLG, based on a CQD system tunnel-coupled to reservoirs, is presented in Figure 1 (a). The two quantum-dots QDα_\alpha (α=a,b\alpha= a, b) are capacitively coupled and interact strongly via Coulomb-interaction [55, 56, 57, 58]. We consider each QDα_\alpha to have a large level spacing, such that only a single energy state is thermally accessible from the reservoirs, with the inter-dot Coulomb-interaction energy UU [59] Figure 1 (b). The dots can thus be either empty (∣0⟩|0\rangle) or occupied (∣1⟩|1\rangle). Repulsive Coulomb-interaction blocks particle exchange between QDs but allows energy transfer via inter-dot coupling [60]. Each QD is tunnel-coupled with one or multiple Fermionic-reservoirs [51, 52, 61], allowing both particle and energy exchange between the reservoir and the QD through the tunnel barrier. The total Hamiltonian for the CQD, baths and their tunneling interactions is given by Htot=Hs+HBath+HTH_{\mathrm{tot}} = H_{\mathrm{s}} + H_{\mathrm{Bath}} + H_{\mathrm{T}}, where Hs=εaNa+εbNb+UNaNbH_{\mathrm{s}} = \varepsilon_{\mathrm{a}}N_{\mathrm{a}} + \varepsilon_{\mathrm{b}}N_{\mathrm{b}} + UN_{\mathrm{a}}N_{\mathrm{b}}. Here, εα\varepsilon_{\alpha} and Nα=dα†dαN_{\alpha} = d_{\alpha}^{\dagger}d_{\alpha} are the single-particle energy and the number operator of the α\alpha-th QD; dα(dα†)d_{\alpha}(d_{\alpha}^{\dagger}) are annihilation (creation) operators, satisfying {dα,dα†}=1\{d_{\alpha},d_{\alpha}^{\dagger}\} = 1 [62, 63, 64]. Thus, the CQD can be seen as a four-level system: ∣A⟩=∣00⟩|A\rangle= |00\rangle, ∣B⟩=∣10⟩|B\rangle= |10\rangle, ∣C⟩=∣01⟩|C\rangle= |01\rangle, and ∣D⟩=∣11⟩|D\rangle= |11\rangle with increasing energies εA=0 μeV\varepsilon_{\mathrm{A}} = 0\ \mu\mathrm{eV}, εB=εa=10 μeV\varepsilon_{\mathrm{B}} = \varepsilon_{\mathrm{a}} = 10\ \mu\mathrm{eV}, εC=εb=15 μeV\varepsilon_{\mathrm{C}} = \varepsilon_{\mathrm{b}} = 15\ \mu\mathrm{eV} and εD=εa+εb+U=37 μeV\varepsilon_{\mathrm{D}} = \varepsilon_{\mathrm{a}} + \varepsilon_{\mathrm{b}} + U = 37\ \mu\mathrm{eV}, respectively Figure 1 (b): Inset. The HBath=∑λHBλH_{\mathrm{Bath}} = \sum_{\lambda} H_{\mathrm{B}}^{\lambda} where HBλ=∑k(ϵk−μλ)ckλ†ckλH_{\mathrm{B}}^{\lambda} = \sum_{k}(\epsilon_{k} - \mu_{\lambda})c_{k}^{\lambda\dagger}c_{k}^{\lambda} and λ={S1,S2,D,I,C}\lambda= \{\mathrm{S1},\mathrm{S2},\mathrm{D},\mathrm{I},\mathrm{C}\}; μλ\mu_{\lambda} is the chemical potential of the λ\lambda-th bath, ϵk\epsilon_{k} being the free-electron energy, and ckλ†(ckλ)c_{k}^{\lambda\dagger}(c_{k}^{\lambda}) are the creation (annihilation) operators of the kk-th bath mode. The tunneling Hamiltonian [49, 51] is HT=∑λαHTλαH_{\mathrm{T}} = \sum_{\lambda\alpha}H_{\mathrm{T}}^{\lambda\alpha}, where, HTλα=ℏ∑k[tkλαckλ†dα+tkλα∗dα†ckλ]H_{\mathrm{T}}^{\lambda\alpha} = \hbar\sum_{k}[t_{k}^{\lambda\alpha}c_{k}^{\lambda\dagger}d_{\alpha} + t_{k}^{\lambda\alpha*}d_{\alpha}^{\dagger}c_{k}^{\lambda}]; with tunneling amplitudes tkλαt_{k}^{\lambda\alpha}. The joint state of the CQD is described by the Lindblad master equation (LME) ddtρs(t)=∑λLλ[ρs(t)]\frac{d}{dt}\rho_{\mathrm{s}}(t) = \sum_{\lambda}\mathcal{L}_{\lambda}[\rho_{\mathrm{s}}(t)] [65, 66], where the Lindbladians Lλ[ρs(t)]\mathcal{L}_{\lambda}[\rho_{\mathrm{s}}(t)] are written for Ohmic-bath [See Appendix A]. At steady state, the LME is a system of four equations with diagonal elements satisfying ρ˙i=0\dot{\rho}_{i} = 0 (i=A,B,C,Di = \mathrm{A}, \mathrm{B}, \mathrm{C}, \mathrm{D}). The LME is used to calculate the steady state heat current (JQλJ_{\mathrm{Q}}^{\lambda}) at the output lead as a function of the temperatures (TλT_{\lambda}) of the input leads, thereby determining the logic operations of the QTLG-circuits. The heat current JQλ=JEλ−μλJNλJ_{\mathrm{Q}}^{\lambda} = J_{\mathrm{E}}^{\lambda} - \mu_{\lambda}J_{\mathrm{N}}^{\lambda}, where energy current JEλ=∑{εij}εijΓijλJ_{\mathrm{E}}^{\lambda} = \sum_{\{\varepsilon_{ij}\}}\varepsilon_{ij}\Gamma_{ij}^{\lambda} and particle current JNλ=∑{εij}ΓijλJ_{\mathrm{N}}^{\lambda} = \sum_{\{\varepsilon_{ij}\}}\Gamma_{ij}^{\lambda} are given in terms of the net-excitation rate Γijλ\Gamma_{ij}^{\lambda} (j>ij > i) induced by λ\lambda-th bath, and the energy difference εij=εj−εi\varepsilon_{ij} = \varepsilon_{j} - \varepsilon_{i} between the CQD states ∣i⟩|i\rangle and ∣j⟩|j\rangle. We note Γijλ=γ[fεijλρi−(1−fεijλ)ρj]=−Γjiλ\Gamma_{ij}^{\lambda} = \gamma[f_{\varepsilon_{ij}}^{\lambda}\rho_{i} - (1 - f_{\varepsilon_{ij}}^{\lambda})\rho_{j}] = -\Gamma_{ji}^{\lambda}, where, γ≃3\gamma\simeq3 GHz [8] is taken as the bare tunneling-rate for all gates and fεijλ=[e(εij−μλ)/kBTλ+1]−1f_{\varepsilon_{ij}}^{\lambda} = [e^{(\varepsilon_{ij}-\mu_{\lambda})/k_{\mathrm{B}}T_{\lambda}} + 1]^{-1} is the Fermi-Dirac (FD) distribution of the bath-λ\lambda [Appendix B].

Schematic of the proposed quantum thermal logic gate and its basic building block

Figure 1. (a) Schematic of the proposed QTLG: the CQD system is tunnel-coupled to S1 (hot/cold, here hot), S2 (hot/cold, here cold), and D (cold) leads; source temperatures define input thermal-logic (hot (red): 1, cold (blue): 0). I and C leads, coupled to QDa_a and QDb_b, are kept at high temperature for NOT and AND gate operations, respectively. The thermal current measured at D (or at I lead for NOT gates) defines the output thermal-logic. (b) The basic building block of a QTLG – a thermal-diode in the forward-bias configuration, where input logic-1 (broadened FD-distribution) allows heat-flow from S to D. Inset: CQD energy levels and its states.

The basic building-block of a QTLG is a CQD placed between a Source (S) and the Drain (D) lead [Fig. 1(b)], forming a thermal-diode [19]. Here, temperatures play a role analogous to electrical voltages, while forward or reverse bias are determined by the position of the Fermi-levels relative to the QD-states. For example, in Figure 1(b), the diode is in forward thermal-bias, with μD\mu_{\mathrm{D}} lying within the occupied and unoccupied states of QDb_{\mathrm{b}}, while μS\mu_{\mathrm{S}} lies below those of QDa_{\mathrm{a}}. Hence, heat flows from the hot S to the cold D (See Appendix C 1 for detailed operation). The opposite alignment corresponds to reverse-bias, where no heat flows from S to D [19], irrespective of the source temperature TST_{\mathrm{S}} [Appendix C 2]. In QTLGs, TST_{\mathrm{S}} of the S-lead(s) defines input-thermal-logic: TS≲TS0T_{\mathrm{S}} \lesssim T_{\mathrm{S}}^{0} (∼50±10\sim50 \pm10 mK), corresponds to logic-0, while TS≳TS1T_{\mathrm{S}} \gtrsim T_{\mathrm{S}}^{1} (∼200±10\sim200 \pm10 mK) corresponds to logic-1. In addition to the S and D, the general QTLG setup includes the Invert (I) and Control (C) leads, coupled to QDa_{\mathrm{a}} and QDb_{\mathrm{b}}, respectively. Both I and C leads are held at high temperature [red in Figure 1(a)], with $T_{\mathrm{I(C)}} \gtrsim 230

mK, analogous to a constant +5V bias in electronic logic-gates [67]. The I-lead enables inverting gate operations (e.g., NOT, NOR, and NAND), whereas the C-lead is employed to obtain AND gate operations. The output thermal-logic is determined by JQλJ_{\mathrm{Q}}^{\lambda} (λ=D\lambda= \mathrm{D} or I) measured at D or I (specifically for inverting gates). If ∣JQD∣,∣JQI∣≲JQ0(∼65±10|J_{\mathrm{Q}}^{\mathrm{D}}|, |J_{\mathrm{Q}}^{\mathrm{I}}| \lesssim J_{\mathrm{Q}}^{0}(\sim65 \pm10 aW): output logic-0; and if ∣JQD∣,∣JQI∣≳JQ1(∼100±10|J_{\mathrm{Q}}^{\mathrm{D}}|, |J_{\mathrm{Q}}^{\mathrm{I}}| \gtrsim J_{\mathrm{Q}}^{1}(\sim100 \pm10 aW): output logic-1. Here, JQDJ_{\mathrm{Q}}^{\mathrm{D}} is negative since it enters the D-lead, while JQIJ_{\mathrm{Q}}^{\mathrm{I}} is positive since it leaves the I-lead. The logic thresholds (JQ0J_{\mathrm{Q}}^{0} and JQ1J_{\mathrm{Q}}^{1}) are chosen based on realistic experimental parameters. A striking one-to-one correspondence between the proposed QTLG and electronic logic-gates becomes apparent as we proceed to discuss the individual gate operations.

Buffer Gate: We begin with the simplest thermal logic gate, the quantum thermal buffer gate (QTBG), in which the output replicates the input logic, and is equivalent to a thermal-diode configured in forward-bias condition. The thermal-circuit of QTBG, shown in Figure 2(a), consists of a single-input (S) coupled to QDa_{\mathrm{a}} and an output (D) coupled to QDb_{\mathrm{b}}. The S can be hot (logic-1) or cold (logic-0), while the D remains cold at the minimum temperature ∼20\sim20 mK (true for all QTLGs). For logic-1, a thermal-bias drives heat from S to D, enforcing unidirectional heat-flow Figure 2(b), analogous to current in an electronic-diode Figure 2(a): Inset.

Figure 2. (a) CQD setup for the QTBG, with its equivalent electrical-diode circuit (Inset). (b) ∣JQD∣|J_{\mathrm{Q}}^{\mathrm{D}}| w.r.t input-temperature TST_{\mathrm{S}}, for forward-bias μD=+21 μeV\mu_{\mathrm{D}} = +21\ \mu\mathrm{eV} and μS=−21 μeV\mu_{\mathrm{S}} = -21\ \mu\mathrm{eV}. (c) QTBG operational cycle for 1→11 \to1 and (d) 0→00 \to0; The net transition-cycle runs in clockwise direction; while the anti-clockwise cycle (⟳) is suppressed, as no hot-lead is coupled to QDb_{\mathrm{b}} [See Appendix-C 2 for details].

To understand the operation of the QTBG, we analyze the heat-flow cycle for input logic-1 Figure 2(c), where TS≳200T_{\mathrm{S}} \gtrsim200 mK. The hot-source is shown by a broadened-FD-distribution (red), while the cold-drain has a sharp FD-distribution (blue). With forward-bias, the cycle proceeds only in clock-wise direction (↻\circlearrowright: A→\toC→\toD→\toB→\toA) and begins with the thermodynamically favorable tunneling of an electron from the D to QDb_{\mathrm{b}} (Appendix C 2), transferring an amount of heat-current equal to εbΓACD\varepsilon_{\mathrm{b}}\Gamma_{\mathrm{AC}}^{\mathrm{D}} and changing the CQD state from ∣A⟩\lvert\mathrm{A}\rangle to ∣C⟩\lvert\mathrm{C}\rangle (panel-1). Due to the Coulomb-interaction, the QDa_{\mathrm{a}}-level shifts to its high-energy state εa+U\varepsilon_{\mathrm{a}}+U (panel-2), requiring an additional amount of energy UU for tunneling. Hot electrons in S (at the tail of the FD) overcome this barrier and tunnel into QDa_{\mathrm{a}}, driving the heat-flow from S and changing the state to ∣D⟩\lvert\mathrm{D}\rangle. This transfer of electrons into QDa_{\mathrm{a}}, in turn, raises the QDb_{\mathrm{b}}-level to (εb+U)(\varepsilon_{\mathrm{b}}+U) (panel-3), allowing the electron in QDb_{\mathrm{b}} to tunnel back to D with an average amount of heat-current −(εb+U)ΓDBD-(\varepsilon_{\mathrm{b}}+U)\Gamma_{\mathrm{DB}}^{\mathrm{D}}, changing the state to ∣B⟩\lvert\mathrm{B}\rangle. Finally, this transition again brings the QDa_{\mathrm{a}}-level back to εa\varepsilon_{\mathrm{a}} (panel-4) and subsequently the electron in QDa_{\mathrm{a}} tunnels back to S, restoring the initial-state ∣A⟩\lvert\mathrm{A}\rangle. The system thus completes a full-cycle, producing a heat current J↻D=−UΓ↻BufJ_{\circlearrowright}^{\mathrm{D}}=-U\Gamma_{\circlearrowright}^{\mathrm{Buf}} entering into D, with steady state net clock-wise cycle-rate Γ↻Buf>0\Gamma_{\circlearrowright}^{\mathrm{Buf}}>0 (Appendix C 2). The output-current ∣JQD∣\lvert J_{\mathrm{Q}}^{\mathrm{D}}\rvert is measurable and significantly higher than JQIJ_{\mathrm{Q}}^{\mathrm{I}}, corresponding to logic-1.

For input logic-0, TS≲50T_{\mathrm{S}}\lesssim50 mK, and the S is described by a sharp FD-distribution (blue), similar to the cold D. The cycle again begins with electron tunneling from D to QDb_{\mathrm{b}} (panel-1), changing the state from ∣00⟩\lvert00\rangle to ∣01⟩\lvert01\rangle and raising the QDa_{\mathrm{a}} level to εa+U\varepsilon_{\mathrm{a}}+U (panel-2). However, electrons in the cold-source lack sufficient thermal energy to overcome the Coulomb-barrier and tunnel into QDa_{\mathrm{a}}. Consequently, the cycle cannot proceed Figure 2(d), and the net heat-current at the drain remains negligibly small, yielding output logic-0. The calculated output current ∣JQD∣\lvert J_{\mathrm{Q}}^{\mathrm{D}}\rvert as a function of TST_{\mathrm{S}} Figure 2(b) clearly displays the two regions of the QTBG truth-table: for TS≲50T_{\mathrm{S}}\lesssim50 mK, ∣JQD∣≪20\lvert J_{\mathrm{Q}}^{\mathrm{D}}\rvert\ll20 aW (0→00\to0); and TS≳200T_{\mathrm{S}}\gtrsim200 mK, ∣JQD∣≫100\lvert J_{\mathrm{Q}}^{\mathrm{D}}\rvert\gg100 aW (1→11\to1).

NOT Gate: A quantum thermal NOT gate (QTNG) can be developed from the QTBG by adding an I-lead to QDa_{\mathrm{a}} Figure 3(a), enabling logic inversion analogous to an electronic NOT gate (Inset). A thermal-bias is applied from S to D, while a constant high-temperature TI≳230T_{\mathrm{I}}\gtrsim230 mK is maintained at the I-lead (similar to constant +5+5V in electronic NOT gate [67]). The thermal current measured in the I-lead defines the output-logic. For a cold-input (TS≲50T_{\mathrm{S}}\lesssim50 mK, logic-0), negligible heat flows from S to D, but the strong thermal-bias at I-lead drives a large output heat-current JQIJ_{\mathrm{Q}}^{\mathrm{I}} to both S and D. This current exceeds threshold, yielding output logic-1 Figure 3(b). For a hot input (TS≳200T_{\mathrm{S}}\gtrsim200 mK, logic-1), both I and S equally contribute to heat-flow to D, lowering JQIJ_{\mathrm{Q}}^{\mathrm{I}} below JQ0∼65J_{\mathrm{Q}}^{0}\sim65 aW and producing output logic-0.

QTNG setup, electrical circuit, output current, and thermal-cycle diagrams

Figure 3. (a) QTNG setup and its electrical circuit (Inset). (b) JQIJ_{\mathrm{Q}}^{\mathrm{I}} w.r.t TST_{\mathrm{S}} for the same parameters as buffer gate, with additional parameter μI=−21\mu_{\mathrm{I}}=-21 μ\mueV, kept below the QDa_{\mathrm{a}} level, maintaining forward-bias. (c) QTNG operational cycle for 0→10\to1 and (d) 1→01\to0. Like in QTBG, the net transition-cycle in ↻\circlearrowright direction; while the anti-clockwise cycle is suppressed (Appendix C 2). Here, the widths of the arrows indicate the relative strength of the heat-flow.

The truth-table for QTNG is illustrated in Figure 3(c,Figure 3d). For thermal input-logic-0 Figure 3(c), the heat-flow cycle begins with tunneling from D to the QDb_{\mathrm{b}} (panel-1), which shifts the QDa_{\mathrm{a}} level to its high-energy state εa+U\varepsilon_{\mathrm{a}}+U (panel-2). The cold S-lead cannot support electron tunneling into the shifted-level εa+U\varepsilon_{\mathrm{a}}+U. However, hot electrons from the I-lead can tunnel into QDa_{\mathrm{a}}, causing a heat-flow and shifting the electron in QDb_{\mathrm{b}} level to the higher energy-state εb+U\varepsilon_{\mathrm{b}}+U (panel-3). Subsequently, the electron in QDb_{\mathrm{b}} tunnels back to D, while the electron in QDa_{\mathrm{a}} mostly tunnels into the S-lead (panel-4), thereby completing the heat-flow cycle. This produces a heat current from the I-lead that exceeds JQIJ_{\mathrm{Q}}^{\mathrm{I}}, implementing the 0→10\to1 operation of the QTNG (See Appendix C3 for details).

For the 1→01\to0 operation, the heat-flow cycle also completes. However, now as S is equally hot as I (input logic-1), both contribute to the heat-cycle and as a result the strength of heat-flow in and out of the I is comparable (indicated by the equal-width arrows to/from I) Figure 3(d), and hence the net output heat-current JQIJ_Q^{I} become negligibly small (<65< 65 aW), corresponding to output logic-0. The calculated JQIJ_Q^{I} versus TST_S is shown in Figure 3(b), from which the NOT gate truth-table follows.

Now, we construct double-input thermal logic gates, beginning with the OR and AND gates and their corresponding NOT operations, NOR and NAND. The proposed thermal-OR gate is shown in Figure 4(a) (top) [inset: electrical equivalent], with two inputs S1, S2, and one output D, forming two thermal-diodes operating in parallel, using each S and the common D [Appendix C § 1.5.4]. If either input satisfies TS1(S2)≳200T_{S1(S2)} \gtrsim200 mK (logic-1), ∣JQD∣≫100\lvert J_Q^{D} \rvert\gg100 aW, giving output logic-1. Only when both inputs satisfy TS1(S2)≲50T_{S1(S2)} \lesssim50 mK (logic-0), ∣JQD∣≪20\lvert J_Q^{D} \rvert\ll20 aW, resulting output logic-0. The calculated ∣JQD∣\lvert J_Q^{D} \rvert in the TS1T_{S1}–TS2T_{S2} plane [Figure 4(a), bottom], generates the thermal-OR gate truth-table.

Double-input QTLG setup and corresponding 2D thermal-current plots for four gates

Figure 4. Double-input QTLG setup and the corresponding 2D thermal-current plot w.r.t. the input T-logic for all four gates. Common parameters: γ=3\gamma= 3 GHz, εa=10\varepsilon_a = 10 μ\mueV, εb=15\varepsilon_b = 15 μ\mueV, U=12U = 12 μ\mueV, TD=20T_D = 20 mK. Additional parameters for (a) OR gate μS1=μS2=−25\mu_{S1} = \mu_{S2} = -25 μ\mueV, μD=21\mu_D = 21 μ\mueV; (b) AND gate: μS1=μS2=16\mu_{S1} = \mu_{S2} = 16 μ\mueV, μC=−72\mu_C = -72 μ\mueV, μD=14\mu_D = 14 μ\mueV; (c) NOR gate: μS1=μS2=−95\mu_{S1} = \mu_{S2} = -95 μ\mueV, μI=−90\mu_I = -90 μ\mueV, μD=21\mu_D = 21 μ\mueV; and (d) NAND gate: μS1=μS2=16\mu_{S1} = \mu_{S2} = 16 μ\mueV, μC=−16\mu_C = -16 μ\mueV, μI=16\mu_I = 16 μ\mueV, μD=9\mu_D = 9 μ\mueV. The equivalent electrical circuit for each gate is shown in the top inset, and output logic states are indicated on the color plots.

A thermal-AND gate is obtained by adding a C-lead with the OR gate [Figure 4(b), top], keeping its temperature fixed at TC≳230T_C \gtrsim230 mK, in analogy to fixed +5V in an electronic AND gate (Inset). The chemical potentials of the leads are chosen in such a way that the two thermal-diodes are connected in reverse-bias w.r.t the cold D, while they are in forward-bias w.r.t the hot C, like in an electronic AND gate [67]. Since a hot C-lead is coupled to QDb_b, the heat-flow cycle can proceed in both directions ↻\circlearrowright and ↺\circlearrowleft (↺\circlearrowleft: A→\toB→\toD→\toC→\toA). When both inputs are 0 (TS1(S2)≲50T_{S1(S2)} \lesssim50 mK), the cold sources cannot populate QDa_a, blocking the ↻\circlearrowright cycle, but ↺\circlearrowleft cycle proceeds with a negligible heat-current below the threshold JQ0J_Q^{0}, giving output logic-0 [Appendix C § 1.5.5 for details]. When one input is 0, and the other is 1, both cycles can proceed; however, due to the position of the μC\mu_C and μD\mu_D, ∣JQD∣\lvert J_Q^{D} \rvert still remains significantly below threshold, and the output logic stays 0 [Appendix C § 1.5.5]. When both input thermal logics are 1, the two hot sources jointly drive the heat-flow, causing ∣JQD∣\lvert J_Q^{D} \rvert to exceed JQ0J_Q^{0}. Consequently, the output logic becomes 1. The color plot of ∣JQD∣\lvert J_Q^{D} \rvert vs TS1T_{S1}, TS2T_{S2} [Figure 4(b), bottom] directly reproduces the truth-table of a thermal-AND gate.

A thermal-NOR gate can be implemented by combining the thermal-OR gate with the thermal-NOT gate [Figure 4(c), top]. Unlike the OR gate, the hot I-lead actively drives the cycle for the (0, 0) case, producing a detectable output (logic-1) when measured at the I-lead. However, if any input is 1, the heat-flow from the I-lead is suppressed, causing the output current to fall well below JQ0J_Q^{0} and yielding output logic-0. All logic operations follow directly from the heat-flow cycle under different input-logic conditions [Appendix C § 1.5.6]. As a hot-lead is attached to QDb_b, the thermal NOR gate also operates via a net ↻\circlearrowright cycle. The calculated JQIJ_Q^{I} [Figure 4(c), bottom] reproduces the complete NOR gate truth-table.

Similar to the NOR gate, a thermal-NAND gate is realized by combining the thermal-AND and thermal-NOT gates [Figure 4(d), top]. Here, unlike the AND gate, if any input is 0, the I-lead actively drives the cycle, and owing to the position of μC\mu_C and μI\mu_I, the output heat current JQIJ_Q^{I} exceeds JQ1J_Q^{1}, giving output logic-1. However, for both input-1, two S-leads and I-lead participate equally in the heat-cycle, and thus JQIJ_Q^{I} is reduced below JQ0J_Q^{0}, resulting in logic-0 [Appendix C § 1.5.7]. The calculated heat-current is plotted in the TS1T_{S1}–TS2T_{S2} plane [Figure 4(d), bottom], producing the complete NAND gate truth-table.

An experimental proposal for the QTLG is presented in Figure 5. The proposed device consists of two QD-junctions, with the nanoparticle QDs in the middle (yellow circles) tunnel-coupled to two or more metallic leads. Here, QDa_a is coupled with three leads, serving as S1 (red), S2 (blue) and I (red), while QDb_b is coupled to two leads, serving as the C (red) and D (blue) leads of the QTLG. Each metallic lead is connected to several superconducting contacts (cyan), which allow for heating (using Joule dissipation) and local thermometry with the use of superconducting junctions [68]. The QD levels can be tuned by the gate voltages Vg1V_{g1} and Vg2V_{g2} to configure the forward/reverse bias condition of thermal-diodes. The input-thermal-logic is determined by the heating voltage (VHV_H) in S1 and S2: logic-1 for VH>0V_H > 0, such that the local-temperature raises above TS1∼200T_S^{1} \sim200 mK, and logic-0 for VH=0V_H = 0, with local-temperature lies below TS0∼50T_S^{0} \sim50 mK. The I and C-leads are heated up to ∼230\sim230 mK using a constant VHV_{\mathrm{H}}, while the D-lead remains cold at the cryostat’s base-temperature (∼10\sim10–2020 mK). With the help of local-thermometers based on superconducting-normal-metal junctions, one can measure the local-temperature of the leads and hence extract the heat-flow rate at the D (or I) and thereby determine the output thermal-logic. Thus, with the same device, one can build several QTLGs by properly selecting the input, output, and control leads. The fabrication of such QD-junctions integrated with local-heater and thermometer is already well-developed in experiments, thanks to the well-established nano-fabrication techniques [14, 39, 44]. The parameters in our calculations are motivated by the currently available experimental data in such CQD-junction devices, e.g., γ∼3\gamma\sim3 GHz, U∼12U \sim12 μ\mueV [8], and lowest measurable heat-current JQ1∼100J_{\mathrm{Q}}^{1} \sim100 aW [44], etc. Thus, the experimental realization of the QTLGs with the proposed device is truly achievable.

Schematic of the proposed device with two QD-junctions, integrated with a local heater and thermometers

Figure 5. Schematic of the proposed device, with two QD-junctions, integrated with a local heater and thermometers (cyan). The QD levels can be tuned by the corresponding back-gates.

In conclusion, the work describes a new concept for implementing thermal logic gates in quantum systems, involving a CQD system tunnel-coupled to Fermionic-leads. The formalism and analysis include realistic experimental parameters, along with an experimental device proposal based on well-established nano-fabrication techniques; thus, the proposed thermal logic device can be realized in future experiments. Moreover, the striking one-to-one correspondence between the QTLGs and conventional classical logic-circuits based on electronic-diodes, in turn, validates the proposal. The proposed device, owing to its components operating in the quantum regime, offers a wide range of opportunities for implementing thermal logic operations in futuristic quantum computation circuits.

We acknowledge Clemens B. Winkelmann for critical reading, comments, and stimulating discussions on the manuscript. SG and AG acknowledge the support from PMRF and IIT Kanpur, respectively. BD acknowledges the support funded by ANRF under the ANRF-PMECRG grant with the project number ANRF/ECRG/2024/003114/PMS, support from the IIT Indore YFRSG grant, and support from PAIR Network of SAKSHAM funded by ANRF, India, through the grant ANRF/PAIR/2025/000018/PAIR, administered and mentored by IIT Indore.

References

Lindblad Master Equation and its Steady-state solution

To derive the quantum master equation for our quantum thermal logic gates (QTLG) model, we start with the von-Neumann equation in the interaction picture for the total density matrix ρtot\rho_{\mathrm{tot}} of the system and bath combined

ddtρtot=−iℏ[HT(t),ρtot(t)].(1)\frac{d}{dt}\rho_{\mathrm{tot}}=-\frac{i}{\hbar}[H_{\mathrm{T}}(t),\rho_{\mathrm{tot}}(t)]. \tag*{(1)}

Here, HTH_{\mathrm{T}} is the tunneling Hamiltonian HT=∑λHTλH_{\mathrm{T}}=\sum_{\lambda}H_{\mathrm{T}}^{\lambda}, where λ={S1,S2,D,I,C}\lambda=\{\mathrm{S1},\mathrm{S2},\mathrm{D},\mathrm{I},\mathrm{C}\}. The HTλH_{\mathrm{T}}^{\lambda} represents the tunneling Hamiltonian between the α\alpha-th (α=a,b\alpha=\mathrm{a},\mathrm{b}) quantum dot (QD), and a λ\lambda-th bath and is given by HTλ≡HTαλ=ℏ∑k[tkαλcλk†cα+tkαλ∗dαk†cλk]H_{\mathrm{T}}^{\lambda}\equiv H_{\mathrm{T}}^{\alpha\lambda}=\hbar\sum_{k}[t_{k}^{\alpha\lambda}c_{\lambda k}^{\dagger}c_{\alpha}+t_{k}^{\alpha\lambda*}d_{\alpha k}^{\dagger}c_{\lambda k}], where tkαλt_{k}^{\alpha\lambda} represents the tunneling amplitude. Integrating the above equation, and taking a trace over the bath degrees of freedom under the Born-Markov approximation [16, 19, 65], one obtains

ρ˙s(t)=1(iℏ)2∑λ,λ′∫0∞dt′ Tr⁡λ[HTλ(t),[HTλ′(t−t′),ρs(t)⊗∏λρλ]],where,λ,λ′={S1,S2,D,I,C}.(2)\dot{\rho}_{\mathrm{s}}(t)=\frac{1}{(i\hbar)^{2}}\sum_{\lambda,\lambda'}\int_{0}^{\infty}dt'\,\operatorname{Tr}_{\lambda}\left[H_{\mathrm{T}}^{\lambda}(t),\left[H_{\mathrm{T}}^{\lambda'}(t-t'),\rho_{\mathrm{s}}(t)\otimes\prod_{\lambda}\rho_{\lambda}\right]\right],\qquad\text{where,}\quad\lambda,\lambda'=\{\mathrm{S1},\mathrm{S2},\mathrm{D},\mathrm{I},\mathrm{C}\}. \tag*{(2)}

Tr⁡λ\operatorname{Tr}_{\lambda} refers to the trace over each bath degree of freedom and Tr⁡λ{ρtot(t)}=ρs(t)\operatorname{Tr}_{\lambda}\{\rho_{\mathrm{tot}}(t)\}=\rho_{\mathrm{s}}(t), being the reduced density operator of the system. Here, ρλ\rho_{\lambda} is the initial density operator of the λ\lambda-th bath which is assumed to be always in equilibrium, and we use the fact that Tr⁡λ{cλ(t)ρλ}=0=Tr⁡λ{cλ†(t)ρλ}\operatorname{Tr}_{\lambda}\{c_{\lambda}(t)\rho_{\lambda}\}=0=\operatorname{Tr}_{\lambda}\{c_{\lambda}^{\dagger}(t)\rho_{\lambda}\} and Tr⁡λ{[HTλ(t),[HTλ′(t−t′),ρs(t)⊗∏λρλ]]}=0; λ≠λ′\operatorname{Tr}_{\lambda}\{[H_{\mathrm{T}}^{\lambda}(t),[H_{\mathrm{T}}^{\lambda'}(t-t'),\rho_{\mathrm{s}}(t)\otimes\prod_{\lambda}\rho_{\lambda}]]\}=0;\ \lambda\ne\lambda'. We further assume that Tr⁡λ[HT(t),ρtot(0)]=0\operatorname{Tr}_{\lambda}[H_{\mathrm{T}}(t),\rho_{\mathrm{tot}}(0)]=0. Finally, eliminating the high-frequency oscillating terms by the standard procedure of secular approximation, we arrive at the Lindblad master equation (LME) of the following form

ρ˙s=∑λLλ[ρs],(3)\dot{\rho}_{\mathrm{s}}=\sum_{\lambda}\mathcal{L}_{\lambda}[\rho_{\mathrm{s}}], \tag*{(3)}

where, the Lindbladian Lλ[ρs]\mathcal{L}_{\lambda}[\rho_{\mathrm{s}}] is given by

Lλ[ρs]=∑{ωα}>0Gλ(ωα)[dα†(ωα)ρsdα(ωα)−12{ρs,dα(ωα)dα†(ωα)}]+Gλ(−ωα)[dα(ωα)ρsdα†(ωα)−12{ρs,dα†(ωα)dα(ωα)}].(4)\mathcal{L}_{\lambda}[\rho_{\mathrm{s}}]=\sum_{\{\omega_{\alpha}\}>0}\mathcal{G}_{\lambda}(\omega_{\alpha})\left[d_{\alpha}^{\dagger}(\omega_{\alpha})\rho_{\mathrm{s}}d_{\alpha}(\omega_{\alpha})-\frac{1}{2}\{\rho_{\mathrm{s}},d_{\alpha}(\omega_{\alpha})d_{\alpha}^{\dagger}(\omega_{\alpha})\}\right]+\mathcal{G}_{\lambda}(-\omega_{\alpha})\left[d_{\alpha}(\omega_{\alpha})\rho_{\mathrm{s}}d_{\alpha}^{\dagger}(\omega_{\alpha})-\frac{1}{2}\{\rho_{\mathrm{s}},d_{\alpha}^{\dagger}(\omega_{\alpha})d_{\alpha}(\omega_{\alpha})\}\right]. \tag*{(4)}

Here {ωα}\{\omega_{\alpha}\} represent the energy associated with the transition between QDα\mathrm{QD}_{\alpha} with its coupled leads. In the present QTLG model, there are four allowed transitions and each lead guides two transitions each. The leads S1, S2, I drive the transitions between ∣A⟩↔∣B⟩|A\rangle\leftrightarrow|B\rangle and ∣D⟩↔∣C⟩|D\rangle\leftrightarrow|C\rangle, while the D, C leads control the ∣A⟩↔∣C⟩|A\rangle\leftrightarrow|C\rangle and ∣D⟩↔∣B⟩|D\rangle\leftrightarrow|B\rangle transitions. To derive the forms of the Lindbladians, we express the creation and annihilation operators in terms of the system eigenstates in the following forms

∑{ωa}da†(ωa)=da†(εAB)+da†(εCD)=∣B⟩⟨A∣+∣D⟩⟨C∣;∑{ωa}da(ωa)=da(εBA)+da(εDC)=∣A⟩⟨B∣+∣C⟩⟨D∣,\sum_{\{\omega_a\}}d_a^{\dagger}(\omega_a)=d_a^{\dagger}(\varepsilon_{AB})+d_a^{\dagger}(\varepsilon_{CD})=|B\rangle\langle A|+|D\rangle\langle C|;\qquad\sum_{\{\omega_a\}}d_a(\omega_a)=d_a(\varepsilon_{BA})+d_a(\varepsilon_{DC})=|A\rangle\langle B|+|C\rangle\langle D|,
∑{ωb}db†(ωb)=db†(εAC)+db†(εBD)=∣C⟩⟨A∣+∣D⟩⟨B∣;∑{ωb}db(ωb)=db(εCA)+db(εDB)=∣A⟩⟨C∣+∣B⟩⟨D∣.(5)\sum_{\{\omega_b\}}d_b^{\dagger}(\omega_b)=d_b^{\dagger}(\varepsilon_{AC})+d_b^{\dagger}(\varepsilon_{BD})=|C\rangle\langle A|+|D\rangle\langle B|;\qquad\sum_{\{\omega_b\}}d_b(\omega_b)=d_b(\varepsilon_{CA})+d_b(\varepsilon_{DB})=|A\rangle\langle C|+|B\rangle\langle D|. \tag*{(5)}

Furthermore, temperature-dependent bath spectral functions in Eq. (4) are defined as

Gλ(ωα)=γλ(ωα)fλ+(ωα);andGλ(−ωα)=γλ(ωα)fλ−(ωα),(6)\mathcal{G}_{\lambda}(\omega_{\alpha})=\gamma_{\lambda}(\omega_{\alpha})f_{\lambda}^{+}(\omega_{\alpha});\qquad\text{and}\qquad\mathcal{G}_{\lambda}(-\omega_{\alpha})=\gamma_{\lambda}(\omega_{\alpha})f_{\lambda}^{-}(\omega_{\alpha}), \tag*{(6)}

where γλ(ωα)\gamma_{\lambda}(\omega_{\alpha}) is the bare tunneling rate between the reservoir λ\lambda and the coupled QDα\mathrm{QD}_{\alpha}, which can be calculated using Fermi’s golden rule, as γλ(εij)=2π∑k∣tkαλ∣2δ(εij−εkλ)\gamma_{\lambda}(\varepsilon_{ij})=2\pi\sum_k|t_{k}^{\alpha\lambda}|^2\delta(\varepsilon_{ij}-\varepsilon_k^{\lambda}) for ωα=εij\omega_{\alpha}=\varepsilon_{ij} and is abbreviated as γλ\gamma_{\lambda} for brevity. For simplicity of our analytical calculations, we later assume that all γλ\gamma_{\lambda} are equal. The fλ±(ωα)f_{\lambda}^{\pm}(\omega_{\alpha}) in Eq. (6) are the Fermi distribution functions calculated via fλ+(ωα=εij)=Tr⁡λ(cλ†cλρλ)f_{\lambda}^{+}(\omega_{\alpha}=\varepsilon_{ij})=\operatorname{Tr}_{\lambda}(c_{\lambda}^{\dagger}c_{\lambda}\rho_{\lambda}) and fλ−(ωα=εij)=Tr⁡λ(cλcλ†ρλ)f_{\lambda}^{-}(\omega_{\alpha}=\varepsilon_{ij})=\operatorname{Tr}_{\lambda}(c_{\lambda}c_{\lambda}^{\dagger}\rho_{\lambda}). The time evolution of the occupation probabilities of the reduced density matrix ρi=⟨i∣ρs(t)∣i⟩\rho_i=\langle i|\rho_{\mathrm{s}}(t)|i\rangle (i=A,B,C,Di=A,B,C,D), is obtained using the LME [Eq. (4)], which leads to a simple loss-gain type rate equation for the steady-state occupation probabilities [65, 66]:

ρ˙A=∑λΓBAλ−∑λΓACλ=0;ρ˙B=∑λΓABλ−∑λΓBDλ=0;\dot{\rho}_{A}=\sum_{\lambda}\Gamma_{BA}^{\lambda}-\sum_{\lambda}\Gamma_{AC}^{\lambda}=0;\qquad\dot{\rho}_{B}=\sum_{\lambda}\Gamma_{AB}^{\lambda}-\sum_{\lambda}\Gamma_{BD}^{\lambda}=0;
ρ˙C=∑λΓDCλ−∑λΓCAλ=0;ρ˙D=∑λΓCDλ−∑λΓDBλ=0.(7)\dot{\rho}_{C}=\sum_{\lambda}\Gamma_{DC}^{\lambda}-\sum_{\lambda}\Gamma_{CA}^{\lambda}=0;\qquad\dot{\rho}_{D}=\sum_{\lambda}\Gamma_{CD}^{\lambda}-\sum_{\lambda}\Gamma_{DB}^{\lambda}=0. \tag*{(7)}

From Eq. (7), we find all Γijλ\Gamma_{ij}^{\lambda} in the steady state are equal and are denoted by clockwise (Γ↻\Gamma_{\circlearrowright}) or anti-clockwise (Γ↺\Gamma_{\circlearrowleft}) transition rates:

∑λΓACλ=∑λΓCDλ=∑λΓDBλ=∑λΓBAλ≡Γ↻;∑λΓABλ=∑λΓBDλ=∑λΓDCλ=∑λΓCAλ≡Γ↺,(8)\sum_{\lambda}\Gamma_{AC}^{\lambda}=\sum_{\lambda}\Gamma_{CD}^{\lambda}=\sum_{\lambda}\Gamma_{DB}^{\lambda}=\sum_{\lambda}\Gamma_{BA}^{\lambda}\equiv\Gamma_{\circlearrowright};\qquad\sum_{\lambda}\Gamma_{AB}^{\lambda}=\sum_{\lambda}\Gamma_{BD}^{\lambda}=\sum_{\lambda}\Gamma_{DC}^{\lambda}=\sum_{\lambda}\Gamma_{CA}^{\lambda}\equiv\Gamma_{\circlearrowleft}, \tag*{(8)}

and satisfies the relation Γ↻=−Γ↺\Gamma_{\circlearrowright}=-\Gamma_{\circlearrowleft}. In Eqs. (A7)–(A8), the net transition rate Γijλ\Gamma_{ij}^{\lambda} from state ∣i⟩→∣j⟩|i\rangle\rightarrow|j\rangle induced by reservoir {λ}\{\lambda\} is given by Γijλ=kijλρi−kjiλρj\Gamma_{ij}^{\lambda}=k_{ij}^{\lambda}\rho_i-k_{ji}^{\lambda}\rho_j; where, kijλ=γλfλ+(ωij)k_{ij}^{\lambda}=\gamma_{\lambda}f_{\lambda}^{+}(\omega_{ij}). The Fermi distribution fλ(εij)f_{\lambda}(\varepsilon_{ij}) associated with the corresponding transition energy εij=εj−εi\varepsilon_{ij}=\varepsilon_j-\varepsilon_i satisfies the following relation fλ(εij)=1−fλ(εji)=[1+exp⁡{βλ(εij−μλ)}]−1f_{\lambda}(\varepsilon_{ij})=1-f_{\lambda}(\varepsilon_{ji})=[1+\exp\{\beta_{\lambda}(\varepsilon_{ij}-\mu_{\lambda})\}]^{-1} [19, 50]; βλ\beta_{\lambda} being the inverse temperature βλ=1/kBTλ\beta_{\lambda}=1/k_{\mathrm{B}}T_{\lambda}. Among the four allowed transitions, εij\varepsilon_{ij} are: εAB=εa\varepsilon_{AB}=\varepsilon_a, εAC=εb\varepsilon_{AC}=\varepsilon_b, εCD=εa+U\varepsilon_{CD}=\varepsilon_a+U, and εBD=εb+U\varepsilon_{BD}=\varepsilon_b+U. For brevity, we denote the Fermi distributions associated with the transition energy εα\varepsilon_{\alpha} as fα1f_{\alpha}^{1}, and fα2f_{\alpha}^{2} with transition energy εα+U\varepsilon_{\alpha}+U. Now, in order to determine the steady-state current, we need the expression of Γ↻\Gamma_{\circlearrowright} or Γ↺\Gamma_{\circlearrowleft}. We rewrite Eq. (7) as

M[ρAρBρCρD]=[0001],(9)\mathcal{M} \begin{bmatrix} \rho_A\\ \rho_B\\ \rho_C\\ \rho_D \end{bmatrix} = \begin{bmatrix} 0\\ 0\\ 0\\ 1 \end{bmatrix}, \tag*{(9)}

where we use Tr⁡{ρ}=1\operatorname{Tr}\{\rho\}=1, i.e., ρA+ρB+ρC+ρD=1\rho_A+\rho_B+\rho_C+\rho_D=1. The matrix M\mathcal{M} is given by

M=[−∑λfλ1∑λfλ1−∑λfλ1−∑λfλ2∑λfλ2∑λfλ10−∑λfλ2−∑λfλ1∑λfλ2∑λfλ1∑λfλ10−∑λfλ2−∑λfλ11111].(10)\mathcal{M}= \begin{bmatrix} -\sum_{\lambda}f_{\lambda}^{1} & \sum_{\lambda}f_{\lambda}^{1} & -\sum_{\lambda}f_{\lambda}^{1}-\sum_{\lambda}f_{\lambda}^{2} & \sum_{\lambda}f_{\lambda}^{2}\\ \sum_{\lambda}f_{\lambda}^{1} & 0 & -\sum_{\lambda}f_{\lambda}^{2}-\sum_{\lambda}f_{\lambda}^{1} & \sum_{\lambda}f_{\lambda}^{2}\\ \sum_{\lambda}f_{\lambda}^{1} & \sum_{\lambda}f_{\lambda}^{1} & 0 & -\sum_{\lambda}f_{\lambda}^{2}-\sum_{\lambda}f_{\lambda}^{1}\\ 1 & 1 & 1 & 1 \end{bmatrix}. \tag*{(10)}

where, we substitute the forms of various Γijλ\Gamma_{ij}^{\lambda} and assume all γλ=γ\gamma_{\lambda}=\gamma. Using Eqs. (9) and A10, one can obtain the populations {ρi}\{\rho_i\} in terms of which one can evaluate the final expression of Γ↻\Gamma_{\circlearrowright} or Γ↺\Gamma_{\circlearrowleft}.

General expressions of steady-state currents

As the composite system weakly interacts with multiple reservoirs and exchanges energy and particles, we derive the expressions for steady-state currents under the grand canonical formalism, following Refs. [66, 69, 70]. Since the system Hamiltonian HsH_{\mathrm{s}} and the total number operator of the system N\mathcal{N} are time-independent, we immediately recognize the net energy flux (JEJ_{\mathrm{E}}) and the particle flux (JNJ_{\mathrm{N}}) as

JE(t)=Tr⁡s[ρ˙s(t)Hs]=∑λTr⁡s[Lλ[ρs(t)]Hs];JN(t)=Tr⁡s[ρ˙s(t)N]=∑λTr⁡s[Lλ[ρs(t)]N],(11)J_{\mathrm{E}}(t) = \operatorname{Tr}_{\mathrm{s}}[\dot{\rho}_{\mathrm{s}}(t)H_{\mathrm{s}}] = \sum_{\lambda}\operatorname{Tr}_{\mathrm{s}}[\mathcal{L}_{\lambda}[\rho_{\mathrm{s}}(t)]H_{\mathrm{s}}]; \qquad J_{\mathrm{N}}(t) = \operatorname{Tr}_{\mathrm{s}}[\dot{\rho}_{\mathrm{s}}(t)\mathcal{N}] = \sum_{\lambda}\operatorname{Tr}_{\mathrm{s}}[\mathcal{L}_{\lambda}[\rho_{\mathrm{s}}(t)]\mathcal{N}], \tag*{(11)}

where we use Eq. (3) for ρ˙s(t)\dot{\rho}_{\mathrm{s}}(t). Identifying JE(t)=∑λJEλ(t)J_{\mathrm{E}}(t) = \sum_{\lambda}J_{\mathrm{E}}^{\lambda}(t) and JN(t)=∑λJNλ(t)J_{\mathrm{N}}(t) = \sum_{\lambda}J_{\mathrm{N}}^{\lambda}(t), we find

JEλ(t)=Tr⁡s[Lλ[ρ(t)]Hs];JNλ(t)=Tr⁡s[Lλ[ρ(t)]N];andJQλ(t)=JEλ(t)−μλJNλ(t),(12)J_{\mathrm{E}}^{\lambda}(t) = \operatorname{Tr}_{\mathrm{s}}[\mathcal{L}_{\lambda}[\rho(t)]H_{\mathrm{s}}]; \qquad J_{\mathrm{N}}^{\lambda}(t) = \operatorname{Tr}_{\mathrm{s}}[\mathcal{L}_{\lambda}[\rho(t)]\mathcal{N}]; \qquad\text{and} \qquad J_{\mathrm{Q}}^{\lambda}(t) = J_{\mathrm{E}}^{\lambda}(t) - \mu_{\lambda}J_{\mathrm{N}}^{\lambda}(t), \tag*{(12)}

where JE(N)λJ_{\mathrm{E(N)}}^{\lambda} or JQλJ_{\mathrm{Q}}^{\lambda} is positive if the energy (particle) or heat current flows from the reservoir λ\lambda to the system. In the steady state, both the energy and the particle currents would be independent of time; thus, the steady-state heat current reduces to

JQλ=JEλ−μλJNλ=Tr⁡s[Lλ[ρss]Hs]−μλTr⁡s[Lλ[ρss]N].(13)J_{\mathrm{Q}}^{\lambda} = J_{\mathrm{E}}^{\lambda} - \mu_{\lambda}J_{\mathrm{N}}^{\lambda} = \operatorname{Tr}_{\mathrm{s}}[\mathcal{L}_{\lambda}[\rho_{\mathrm{ss}}]H_{\mathrm{s}}] - \mu_{\lambda}\operatorname{Tr}_{\mathrm{s}}[\mathcal{L}_{\lambda}[\rho_{\mathrm{ss}}]\mathcal{N}]. \tag*{(13)}

Using Eq. (4) we obtain the formal expressions for the energy, particle, and heat currents as

JEλ=∑{ϵij}ϵijΓijλ;JNλ=∑{ϵij}Γijλ;JQλ=∑{ϵij}(ϵij−μλ)Γijλ.(14)J_{\mathrm{E}}^{\lambda} = \sum_{\{\epsilon_{ij}\}}\epsilon_{ij}\Gamma_{ij}^{\lambda}; \qquad J_{\mathrm{N}}^{\lambda} = \sum_{\{\epsilon_{ij}\}}\Gamma_{ij}^{\lambda}; \qquad J_{\mathrm{Q}}^{\lambda} = \sum_{\{\epsilon_{ij}\}}(\epsilon_{ij} - \mu_{\lambda})\Gamma_{ij}^{\lambda}. \tag*{(14)}

where, we note that ϵij=−ϵji\epsilon_{ij} = -\epsilon_{ji} and Γijλ=−Γjiλ\Gamma_{ij}^{\lambda} = -\Gamma_{ji}^{\lambda}.

Steady-state transition rates and currents for all gates

Detailed Operation of thermal diode

Let us now describe the detailed operation of the thermal diode shown in Figure 1(b) of the main text. Here, two QDs, QDa_{\mathrm{a}} and QDb_{\mathrm{b}}, are tunnel-coupled to the Source (S) and Drain (D) leads, respectively. Under forward-bias conditions, heat flows from the hot S lead to the cold D lead, whereas under reverse-bias conditions, no heat flow occurs [see Fig. 6].

In the above figure, only the clockwise transition cycles are shown for both forward and reverse bias configurations, while the anticlockwise transition cycle doesn’t proceed and is not shown. The reason for this is explained in the following section through the concept of a buffer gate, where we show that the thermal diode can function as a buffer gate.

Thermal diode as Buffer Gate

For the above diode model, the steady-state transition rates for the clockwise and anti-clockwise transition cycles are illustrated via a diagrammatic picture in Figure 7. According to Eq. (8), the transition rates satisfy the following relations

Figure 7. Diagrammatic representation for clockwise and anticlockwise transition cycles for buffer gate.

Figure 6. Operation of thermal diode: (Left) In the forward-bias configuration, heat flows from S to D when TS>TDT_S > T_D, but no heat flows from D to S even when TD>TST_D > T_S, i.e., only a unidirectional heat-flow is allowed in a thermal diode. (Right) In the reverse-bias configuration, no heat flows regardless of whether TST_S is greater than or less than TDT_D. In both cases, the cycles are shown in the ↻\circlearrowright direction, corresponding to the thermodynamically preferred direction of heat flow.

ΓACD=ΓCDS=ΓDBD=ΓBAS≡Γ↻Buf ;ΓABS=ΓBDD=ΓDCS=ΓCAD≡Γ↺Buf ;(15)\Gamma_{\mathrm{AC}}^{\mathrm{D}}=\Gamma_{\mathrm{CD}}^{\mathrm{S}}=\Gamma_{\mathrm{DB}}^{\mathrm{D}}=\Gamma_{\mathrm{BA}}^{\mathrm{S}}\equiv\Gamma_{\circlearrowright}^{\mathrm{Buf}} \ ;\quad\Gamma_{\mathrm{AB}}^{\mathrm{S}}=\Gamma_{\mathrm{BD}}^{\mathrm{D}}=\Gamma_{\mathrm{DC}}^{\mathrm{S}}=\Gamma_{\mathrm{CA}}^{\mathrm{D}}\equiv\Gamma_{\circlearrowleft}^{\mathrm{Buf}} \ ; \tag*{(15)}

where

Γ↻Buf=−Γ↺Buf=γ2[fD2fD2fS2−fD2fD2fS1+fS1fS2fD1−fS1fS2fD2+fS1fD2−fS2fD2−fS2fD1fS1fD+fS2fD2−fS1fD2−fS2fD1−1].(16)\Gamma_{\circlearrowright}^{\mathrm{Buf}}=-\Gamma_{\circlearrowleft}^{\mathrm{Buf}}=\frac{\gamma}{2}\left[\frac{f_{\mathrm{D}}^{2}f_{\mathrm{D}}^{2}f_{\mathrm{S}}^{2}-f_{\mathrm{D}}^{2}f_{\mathrm{D}}^{2}f_{\mathrm{S}}^{1}+f_{\mathrm{S}}^{1}f_{\mathrm{S}}^{2}f_{\mathrm{D}}^{1}-f_{\mathrm{S}}^{1}f_{\mathrm{S}}^{2}f_{\mathrm{D}}^{2}+f_{\mathrm{S}}^{1}f_{\mathrm{D}}^{2}-f_{\mathrm{S}}^{2}f_{\mathrm{D}}^{2}-f_{\mathrm{S}}^{2}f_{\mathrm{D}}^{1}}{f_{\mathrm{S}}^{1}f_{\mathrm{D}}+f_{\mathrm{S}}^{2}f_{\mathrm{D}}^{2}-f_{\mathrm{S}}^{1}f_{\mathrm{D}}^{2}-f_{\mathrm{S}}^{2}f_{\mathrm{D}}^{1}-1}\right]. \tag*{(16)}

Since the output current for the buffer gate is measured in the drain lead, substituting (15) into the general expressions for the steady-state currents given by (14), we find the energy and particle currents at the drain lead are:

JED=εbΓACD+[−(εb+U)]ΓDBD=εbΓ↻Buf−(εb+U)Γ↻Buf=−UΓ↻Buf,(17)\begin{aligned} J_{\mathrm{E}}^{\mathrm{D}}=\varepsilon_{\mathrm{b}}\Gamma_{\mathrm{AC}}^{\mathrm{D}}+\left[-(\varepsilon_{\mathrm{b}}+U)\right]\Gamma_{\mathrm{DB}}^{\mathrm{D}} \\ &=\varepsilon_{\mathrm{b}}\Gamma_{\circlearrowright}^{\mathrm{Buf}}-(\varepsilon_{\mathrm{b}}+U)\Gamma_{\circlearrowright}^{\mathrm{Buf}} \\ &=-U\Gamma_{\circlearrowright}^{\mathrm{Buf}}, \tag*{(17)} \end{aligned}

and

JND=ΓACD+ΓBDD=0,(18)J_{\mathrm{N}}^{\mathrm{D}}=\Gamma_{\mathrm{AC}}^{\mathrm{D}}+\Gamma_{\mathrm{BD}}^{\mathrm{D}}=0, \tag*{(18)}

respectively. Since inter-dot particle hopping is prohibited, JND=0J_{\mathrm{N}}^{\mathrm{D}}=0. Consequently, JQDJ_{\mathrm{Q}}^{\mathrm{D}} becomes

JQD=JED−μDJND=−UΓ↻Buf<0,(19)J_{\mathrm{Q}}^{\mathrm{D}}=J_{\mathrm{E}}^{\mathrm{D}}-\mu_{\mathrm{D}}J_{\mathrm{N}}^{\mathrm{D}}=-U\Gamma_{\circlearrowright}^{\mathrm{Buf}}<0, \tag*{(19)}

where Γ↻Buf\Gamma_{\circlearrowright}^{\mathrm{Buf}} is given by (16). Here, due to the temperature gradient between S and D with TS>TDT_S>T_D, the heat-current JQDJ_{\mathrm{Q}}^{\mathrm{D}} enters into the D-lead, thus the JQDJ_{\mathrm{Q}}^{\mathrm{D}} is negative. Therefore, the clockwise cycle rate Γ↻Buf\Gamma_{\circlearrowright}^{\mathrm{Buf}} must be positive for the heat to flow from the hot source to cold D, favorable by laws of thermodynamics. Since Γ↺Buf=−Γ↻Buf\Gamma_{\circlearrowleft}^{\mathrm{Buf}}=-\Gamma_{\circlearrowright}^{\mathrm{Buf}}, this further implies that Γ↺Buf<0\Gamma_{\circlearrowleft}^{\mathrm{Buf}}<0 and hence the heat-current at D in the anticlockwise cycle is positive, which indicates the heat-flow direction from cold D to hot S. However, this violates the laws of thermodynamics. Therefore, the anticlockwise cycle (with Γ↺Buf<0\Gamma_{\circlearrowleft}^{\mathrm{Buf}}<0) is highly improbable on average. As a result, for the buffer gate, only the clockwise cycle operates in accordance with the laws of thermodynamics, with a net clockwise cycle rate Γ↻Buf>0\Gamma^{\mathrm{Buf}}_{\circlearrowright} > 0. As a result, it is equivalent to a thermal diode which operates only under forward bias conditions, where heat flows from hot S to cold D lead, and remains inactive under reverse bias [Fig. 6].

Figure 8. Thermal Buffer and NOT gate anticlockwise cycle: (a) and (b) correspond to the buffer gate with logic inputs 1 and 0, respectively, while (c) and (d) correspond to the NOT gate with logic inputs 0 and 1, respectively. In cases (a), (c), and (d), the drain lead does not provide sufficient thermal energy to excite an electron into the excited state of QDb_b (i.e., ΓBDD≈0\Gamma^{\mathrm{D}}_{\mathrm{BD}} \approx0); hence, the cycle cannot be completed. In case (b), since the source (S) lead is cold, it does not have sufficient thermal energy to initiate the anti-clockwise cycle (i.e., ΓABS≈0\Gamma^{\mathrm{S}}_{\mathrm{AB}} \approx0); therefore, none of the anti-clockwise cycles operate.

As discussed above, the buffer gate operates only through the clockwise cycle under forward bias. This behavior is not restricted to the buffer gate only but also applies to the NOT, OR, and NOR gates. In all these cases, the QDb_b is coupled only to a cold drain lead, maintained at the lowest temperature. In the anti-clockwise cycle, the cycle initiates with the tunneling from S to QDa_a, which raises the QDb_b level to its high energy state (εb+U\varepsilon_b + U). However, the cold drain can not provide sufficient energy to overcome this barrier and tunnel electrons into QDb_b; as a result, the cycle can not proceed further to produce any heat current [Fig. 8]. This remains valid as long as QDb_b is coupled only to a single cold D lead. Accordingly, in the discussion of NOT, OR, and NOR gates, we consider only the clockwise transition cycle in both the diagrammatic representation in the Appendix and the main text. However, for the sake of completeness, we also provide the anticlockwise cycles for all four above gates [see Figs. 8, 12, and 18]. This situation, however, changes for the AND and NAND gates, where QDb_b is additionally coupled to a hot control (C) lead maintained at a fixed high temperature. In this case, both cycles can, in principle, occur, and we discuss them in detail in the analysis of the AND and NAND gate operations.

Thermal-NOT Gate

By coupling an additional Invert lead (I) to QDa_a, characterized by temperature TIT_{\mathrm{I}} and chemical potential μI\mu_{\mathrm{I}}, we control the output thermal current JQJ_{\mathrm{Q}} at the I lead. This upgrades the thermal-diode model to a thermal-NOT gate: QDa_a is coupled to the source (S) and invert (I) leads, while QDb_b is coupled to the drain lead D; therefore, applying the same analysis as before, we obtain [Fig. 9]:

ΓACD=ΓCDSI=ΓDBD=ΓBASI=Γ↻NOT(20)\Gamma^{\mathrm{D}}_{\mathrm{AC}} = \Gamma^{\mathrm{SI}}_{\mathrm{CD}} = \Gamma^{\mathrm{D}}_{\mathrm{DB}} = \Gamma^{\mathrm{SI}}_{\mathrm{BA}} = \Gamma^{\mathrm{NOT}}_{\circlearrowright} \tag*{(20)}

where ΓijSI≡ΓijS+ΓijI\Gamma^{\mathrm{SI}}_{ij} \equiv\Gamma^{\mathrm{S}}_{ij} + \Gamma^{\mathrm{I}}_{ij}.

Note that Figure 9 is very similar to Figure 7 and shows the clockwise cycle operating under forward bias. However, the only difference is that the ∣A⟩↔∣B⟩\lvert A\rangle\leftrightarrow\lvert B\rangle and ∣C⟩↔∣D⟩\lvert C\rangle\leftrightarrow\lvert D\rangle transitions are now controlled by both the S and I leads. As a result, when we use the above relations in Eq. (14), the steady-state energy and particle currents associated with the I lead take the following form

Figure 9. Diagrammatic representation for the clockwise transition cycle for the NOT gate.

JEI=εaΓABI+(εa+U)ΓCDI;JNI=ΓABI+ΓCDI.(21)J_{\mathrm{E}}^{\mathrm{I}}=\varepsilon_{\mathrm{a}}\Gamma_{\mathrm{AB}}^{\mathrm{I}}+(\varepsilon_{\mathrm{a}}+U)\Gamma_{\mathrm{CD}}^{\mathrm{I}};\qquad J_{\mathrm{N}}^{\mathrm{I}}=\Gamma_{\mathrm{AB}}^{\mathrm{I}}+\Gamma_{\mathrm{CD}}^{\mathrm{I}}. \tag*{(21)}

Therefore, the heat current associated with the I lead reads,

JQI=JEI−μIJNI=(εa−μI)ΓABI+(εa+U−μI)ΓCDI=(εa−μI)(ΓCDI−ΓBAI)+UΓCDI.(22)J_{\mathrm{Q}}^{\mathrm{I}}=J_{\mathrm{E}}^{\mathrm{I}}-\mu_{\mathrm{I}}J_{\mathrm{N}}^{\mathrm{I}}=(\varepsilon_{\mathrm{a}}-\mu_{\mathrm{I}})\Gamma_{\mathrm{AB}}^{\mathrm{I}}+(\varepsilon_{\mathrm{a}}+U-\mu_{\mathrm{I}})\Gamma_{\mathrm{CD}}^{\mathrm{I}}=(\varepsilon_{\mathrm{a}}-\mu_{\mathrm{I}})(\Gamma_{\mathrm{CD}}^{\mathrm{I}}-\Gamma_{\mathrm{BA}}^{\mathrm{I}})+U\Gamma_{\mathrm{CD}}^{\mathrm{I}}. \tag*{(22)}

The output logic value of the thermal NOT gate is determined through the magnitude of the heat current JQIJ_{\mathrm{Q}}^{\mathrm{I}}, obtained from the above analytical expression, where Γijλ=kijλρi−kjiλρj\Gamma_{ij}^{\lambda}=k_{ij}^{\lambda}\rho_i-k_{ji}^{\lambda}\rho_j and kijλ=γfλ±(ωij)k_{ij}^{\lambda}=\gamma f_{\lambda}^{\pm}(\omega_{ij}).

  • For input logic 0, it is evident from the transition cycle of the NOT gate [Figure 3(c) of the main text] that ΓCDI>ΓBAI\Gamma_{\mathrm{CD}}^{\mathrm{I}}>\Gamma_{\mathrm{BA}}^{\mathrm{I}}, since the transition from ∣C⟩\lvert C\rangle to ∣D⟩\lvert D\rangle is predominantly driven by the I lead, whereas the de-excitation process from ∣B⟩\lvert B\rangle to ∣A⟩\lvert A\rangle is shared between the I and S leads, where cold S lead dominates the de-excitation process. Furthermore, the chosen parameters satisfy μI<0\mu_{\mathrm{I}}<0 and εa,U>0\varepsilon_{\mathrm{a}},U>0. Substituting these parameters into Eq. (22), it follows that the output current not only satisfies JQI>0J_{\mathrm{Q}}^{\mathrm{I}}>0, but also it exceeds the threshold value JQ1J_{\mathrm{Q}}^{1} and therefore corresponds to the output logic-1.

  • For input logic 1, the transition cycle of the NOT gate [Figure 3(d) of the main text] indicates that ΓCDI\Gamma_{\mathrm{CD}}^{\mathrm{I}} decreases (compared to the logic-0 case), since under this condition the excitation process from ∣C⟩\lvert C\rangle to ∣D⟩\lvert D\rangle also receives contribution from the hot S lead. In contrast, ΓBAI\Gamma_{\mathrm{BA}}^{\mathrm{I}} increases (compared to the logic-0 case) due to the absence of cold S lead. According to Eq. (22), these conditions drive the output current JQIJ_{\mathrm{Q}}^{\mathrm{I}} below the threshold value JQ1J_{\mathrm{Q}}^{1}, corresponding to the output logic-0.

Thermal-OR Gate

The thermal-OR gate is a two-input gate, so we couple two source leads S1 and S2 to QDa_{\mathrm{a}}. QDb_{\mathrm{b}} is tunnel-coupled to a single lead D. The input logic is defined by the temperature of the S1 and S2 leads, whereas the magnitude of the heat current ∣JQD∣\lvert J_{\mathrm{Q}}^{\mathrm{D}}\rvert measured at the drain lead determines the output logic. At the steady-state, all transition rates ∑λΓijλ\sum_{\lambda}\Gamma_{ij}^{\lambda} [Figure 10] associated with the transition ∣i⟩↔∣j⟩\lvert i\rangle\leftrightarrow\lvert j\rangle are equal [cf. Eq. (8)]. For the OR gate, this results in the following expressions of the transition rates under a clockwise cycle:

Figure 10. Diagrammatic representation for the clockwise transition cycle for the OR gate.

ΓACD=ΓCDS1S2=ΓDBD=ΓBAS1S2=Γ↻OR(23)\Gamma_{\mathrm{AC}}^{\mathrm{D}}=\Gamma_{\mathrm{CD}}^{\mathrm{S1S2}}=\Gamma_{\mathrm{DB}}^{\mathrm{D}}=\Gamma_{\mathrm{BA}}^{\mathrm{S1S2}}=\Gamma_{\circlearrowright}^{\mathrm{OR}} \tag*{(23)}

where, ΓijS1S2≡ΓijS1+ΓijS2\Gamma_{ij}^{\mathrm{S1S2}}\equiv\Gamma_{ij}^{\mathrm{S1}}+\Gamma_{ij}^{\mathrm{S2}}.

Similar to the buffer gate, the steady-state energy and particle currents at the drain lead are obtained upon inserting Eq. (23) into Eq. (14), and given by

JED=εbΓACD+(εb+U)ΓBDD=εbΓ↻OR−(εb+U)Γ↻OR=−UΓ↻OR,(24)J_{\mathrm{E}}^{\mathrm{D}}=\varepsilon_{\mathrm{b}}\Gamma_{\mathrm{AC}}^{\mathrm{D}}+(\varepsilon_{\mathrm{b}}+U)\Gamma_{\mathrm{BD}}^{\mathrm{D}}=\varepsilon_{\mathrm{b}}\Gamma_{\circlearrowright}^{\mathrm{OR}}-(\varepsilon_{\mathrm{b}}+U)\Gamma_{\circlearrowright}^{\mathrm{OR}}=-U\Gamma_{\circlearrowright}^{\mathrm{OR}}, \tag*{(24)}

and

JND=ΓACD+ΓBDD=0.(25)J_{\mathrm{N}}^{\mathrm{D}}=\Gamma_{\mathrm{AC}}^{\mathrm{D}}+\Gamma_{\mathrm{BD}}^{\mathrm{D}}=0. \tag*{(25)}

Hence, the heat current associated with the drain lead follows

JQD=JED−μDJND=−UΓ↻OR.(26)J_{\mathrm{Q}}^{\mathrm{D}}=J_{\mathrm{E}}^{\mathrm{D}}-\mu_{\mathrm{D}}J_{\mathrm{N}}^{\mathrm{D}}=-U\Gamma_{\circlearrowright}^{\mathrm{OR}}. \tag*{(26)}

Figure 11. Thermal-OR gate heat-flow cycle is equivalent to two thermal-diodes operating in parallel, formed out of each S and the common D. For input logic (0, 0), both S leads are cold and can not provide sufficient energy for the electrons to tunnel into QDa_a, and thus the heat-flow cycle can not proceed, like in QTBGs, yielding thermal output logic-0. While for (1, 0), (0, 1), and (1, 1), at least one of the S leads is hot enough to tunnel electrons and drive the ⟳ cycle under forward-bias, thereby yielding output logic-1. Parameters are: μS1=μS2=−25 μeV\mu_{\mathrm{S1}} = \mu_{\mathrm{S2}} = -25\ \mu\mathrm{eV}, and μD=21 μeV\mu_{\mathrm{D}} = 21\ \mu\mathrm{eV}.

where

Γ↻OR=−Γ↺OR=γ[fS1S21[fS1S22(fD2−fD1)+2fD2(fD1−1)]−2fS1S21fS1S22(fD2−1)3fS1S21(fD1−fD2)−6+3fS1S22(fD2−fD1)].(27)\Gamma_{\circlearrowright}^{\mathrm{OR}}=-\Gamma_{\circlearrowleft}^{\mathrm{OR}}=\gamma\left[\frac{f_{\mathrm{S1S2}}^{1}\left[f_{\mathrm{S1S2}}^{2}\left(f_{\mathrm{D}}^{2}-f_{\mathrm{D}}^{1}\right)+2f_{\mathrm{D}}^{2}\left(f_{\mathrm{D}}^{1}-1\right)\right]-2f_{\mathrm{S1S2}}^{1}f_{\mathrm{S1S2}}^{2}\left(f_{\mathrm{D}}^{2}-1\right)}{3f_{\mathrm{S1S2}}^{1}\left(f_{\mathrm{D}}^{1}-f_{\mathrm{D}}^{2}\right)-6+3f_{\mathrm{S1S2}}^{2}\left(f_{\mathrm{D}}^{2}-f_{\mathrm{D}}^{1}\right)}\right]. \tag*{(27)}

where, fS1S21(2)=fS11(2)+fS21(2)f_{\mathrm{S1S2}}^{1(2)}=f_{\mathrm{S1}}^{1(2)}+f_{\mathrm{S2}}^{1(2)}. The magnitude of JQDJ_{\mathrm{Q}}^{\mathrm{D}} determines the output logic value: if it is above the JQ0∼100 aWJ_{\mathrm{Q}}^{0}\sim100\ \mathrm{aW}, logic is 1, and if it is less than JQ0∼65 aWJ_{\mathrm{Q}}^{0}\sim65\ \mathrm{aW}, logic is 0.

  • For input logic (0,0), it is evident that the transition cycle becomes frozen, since the transition from ∣C⟩\lvert\mathrm{C}\rangle to ∣D⟩\lvert\mathrm{D}\rangle is not driven by either of the source leads due to the absence of hot electrons [Fig. 11]. Consequently, the steady-state transition rate Γ↻OR\Gamma_{\circlearrowright}^{\mathrm{OR}} approaches zero. Therefore, from Eq. (26), the output current ∣JQD∣\lvert J_{\mathrm{Q}}^{\mathrm{D}}\rvert becomes negligible, corresponding to the output logic 0.

  • For input logic (1,0) or (0,1), the transition cycle proceeds in the clockwise direction, since the transition from ∣C⟩\lvert\mathrm{C}\rangle to ∣D⟩\lvert\mathrm{D}\rangle is facilitated by the hot electron originating from one of the hot S leads [Fig. 11]. As a result, a significant non-zero value of Γ↻OR\Gamma_{\circlearrowright}^{\mathrm{OR}} is generated. Substituting this into Eq. (26), we obtain a considerable output heat current ∣JQD∣\lvert J_{\mathrm{Q}}^{\mathrm{D}}\rvert that exceeds the threshold value JQ1J_{\mathrm{Q}}^{1}, thereby corresponding to the output logic 1.

  • For input logic (1,1), it indicates a significant enhancement of the steady-state transition rate Γ↻OR\Gamma_{\circlearrowright}^{\mathrm{OR}}. This increase arises from the participation of both hot S leads in facilitating the transition from ∣C⟩\lvert\mathrm{C}\rangle to ∣D⟩\lvert\mathrm{D}\rangle [Fig. 11]. Consequently, as follows from Eq. (26), the output heat current ∣JQD∣\lvert J_{\mathrm{Q}}^{\mathrm{D}}\rvert also increases substantially and remains well above the threshold value JQ1J_{\mathrm{Q}}^{1}, thereby corresponding to the output logic 1.

Thermal-AND Gate

The thermal AND gate can be constructed from the OR-gate by introducing an additional control (C) lead attached to QDb_b. In this configuration, QDa_a is tunnel-coupled to the source leads S1 and S2. In contrast, QDb_b is tunnel-coupled to one drain lead D and a control lead C. Under these conditions, the steady-state transition rates [Fig. 13] obey the following relation

Figure 12. Thermal-OR gate anticlockwise cycle: Since no hot lead is attached to QDb_b, the cold D lead cannot supply sufficient thermal energy to pump an electron into QDb_b; thus, ΓBDD≈0\Gamma^{D}_{BD} \approx0, and the anti-clockwise cycle cannot be completed. For the input state (0,0), the anti-clockwise cycle cannot even be initiated because both S leads are cold (i.e., ΓABS1S2≈0\Gamma^{S1S2}_{AB} \approx0). Parameters for OR gate are: μS1=μS2=−25 μeV\mu_{S1} = \mu_{S2} = -25~\mu\mathrm{eV}, and μD=21 μeV\mu_D = 21~\mu\mathrm{eV}.

ΓACDC=ΓCDS1S2=ΓDBDC=ΓBAS1S2≡Γ↻AND ;ΓABS1S2=ΓBDDC=ΓDCS1S2=ΓCADC≡Γ↺AND(28)\Gamma^{DC}_{AC} = \Gamma^{S1S2}_{CD} = \Gamma^{DC}_{DB} = \Gamma^{S1S2}_{BA} \equiv\Gamma^{\mathrm{AND}}_{\circlearrowright}\ ;\quad\Gamma^{S1S2}_{AB} = \Gamma^{DC}_{BD} = \Gamma^{S1S2}_{DC} = \Gamma^{DC}_{CA} \equiv\Gamma^{\mathrm{AND}}_{\circlearrowleft} \tag*{(28)}

where ΓijS1S2≡ΓijS1+ΓijS2\Gamma^{S1S2}_{ij} \equiv\Gamma^{S1}_{ij} + \Gamma^{S2}_{ij} and ΓijDC≡ΓijD+ΓijC\Gamma^{DC}_{ij} \equiv\Gamma^{D}_{ij} + \Gamma^{C}_{ij}, with Γ↻AND=−Γ↺AND\Gamma^{\mathrm{AND}}_{\circlearrowright} = -\Gamma^{\mathrm{AND}}_{\circlearrowleft}. Since the QDb_b is coupled to both cold (D) and hot (C) leads, the transition cycle can run in both clockwise and anti-clockwise directions Figure 13. Like the buffer and OR gate, the AND gate output is measured at the D-lead. So, substituting the above relations into Eq. (14), the steady-state energy and particle currents associated with the drain lead are obtained as

Figure 13. Diagrammatic representation for clockwise and anticlockwise transition cycles for the AND gate.

JED=εbΓACD+(εb+U)ΓBDD;JND=ΓACD+ΓBDD.(29)J^{D}_{E} = \varepsilon_b\Gamma^{D}_{AC} + (\varepsilon_b + U)\Gamma^{D}_{BD};\qquad J^{D}_{N} = \Gamma^{D}_{AC} + \Gamma^{D}_{BD}. \tag*{(29)}

Consequently, the heat current measured at the drain lead can be calculated as

JQD=(εb−μD)ΓACD+(εb+U−μD)ΓBDD=(εb−μD)(ΓACD−ΓDBD)−UΓDBD=(εb−μD)(ΓBDD−ΓCAD)+UΓBDD.(30)\begin{aligned} J^{D}_{Q} &= (\varepsilon_b - \mu_D)\Gamma^{D}_{AC} + (\varepsilon_b + U - \mu_D)\Gamma^{D}_{BD}\\ &= (\varepsilon_b - \mu_D)(\Gamma^{D}_{AC} - \Gamma^{D}_{DB}) - U\Gamma^{D}_{DB}\\ &= (\varepsilon_b - \mu_D)(\Gamma^{D}_{BD} - \Gamma^{D}_{CA}) + U\Gamma^{D}_{BD}. \tag*{(30)} \end{aligned}

The output logic of the thermal AND gate is determined by the magnitude of ∣JDQ∣|J_{\mathrm{D}}^{\mathrm{Q}}|, obtained from the above analytical expression, while the temperatures of the S1 and S2 leads define the thermal input logic. Unlike the previously discussed gates, both clockwise and anti-clockwise transition cycles can occur in the AND gate configuration [Fig. 15], since QDb\mathrm{QD}_{\mathrm{b}} is coupled to the hot C lead. Consequently, the logical operation of this gate is comparatively more intricate, as the contributions from both transition cycles must be taken into account. However, the crucial contribution arises from the participation of the drain lead in the transitions between ∣B⟩|\mathrm{B}\rangle and ∣D⟩|\mathrm{D}\rangle, and between ∣C⟩|\mathrm{C}\rangle and ∣A⟩|\mathrm{A}\rangle, as indicated by Eq. (30). Since the configurations of the C and D leads remain unchanged with varying input logic, a finite heat current is always driven from the hot C lead to the cold D lead. For the clockwise transition cycle [Fig. 14], the transition from ∣A⟩|\mathrm{A}\rangle to ∣C⟩|\mathrm{C}\rangle is predominantly driven by the C lead, whereas both leads participate in the transition from ∣D⟩|\mathrm{D}\rangle to ∣B⟩|\mathrm{B}\rangle. Consequently, a significant amount of heat current flows into the drain lead during the clockwise cycle. In contrast, for the anti-clockwise transition cycle [Fig. 15], the C lead must facilitate the transition between ∣B⟩|\mathrm{B}\rangle and ∣D⟩|\mathrm{D}\rangle, which requires more energy (as μC\mu_{\mathrm{C}} is placed much below QDbQ_{\mathrm{D}_{\mathrm{b}}} level) and is therefore less favorable than the excitation between ∣A⟩|\mathrm{A}\rangle and ∣C⟩|\mathrm{C}\rangle (in the clockwise cycle), which has a lower energy cost. Moreover, both leads contribute to the de-excitation transition from ∣C⟩|\mathrm{C}\rangle to ∣A⟩|\mathrm{A}\rangle. Therefore, the anti-clockwise cycle [Fig. 15] generates a comparatively smaller heat current in the drain lead than the clockwise cycle.

Thus, it is evident that for both transition cycles, the heat current flows into the drain lead. Hence, the simultaneous occurrence of both cycles does not cancel the drain heat current; rather, it enhances the overall heat flow. However, the magnitude of the output heat current contribution from the clockwise transition cycle becomes comparatively larger, whereas the anti-clockwise cycle contributes only a small amount (which remains below the threshold JQJ_{\mathrm{Q}}) to the overall current.

  • For input logic (0,0), the clockwise transition cycle [Fig. 14] becomes frozen, since the transition from

Figure 14. Thermal-AND gate heat-flow cycle can proceed in both directions; clockwise cycle is shown here (cf. Figure 15 for anti-clockwise cycle). The chemical potentials are tuned to keep both thermal-diodes reverse-bias w.r.t the cold D and forward-bias w.r.t the hot C. For (0,0), clockwise cycle is blocked, although it yields negligible heat-current due to anti-clockwise cycle with net output logic-0. For (1,0), both cycles proceed, but position of μC\mu_{\mathrm{C}} and μD\mu_{\mathrm{D}}, suppress the tunnel rates and the heat-current (thin arrow), yielding logic-0. For both inputs 1, both sources drive the cycle, and the net output exceeds the threshold (thick arrow), yielding logic-1. Parameters are: μS1=μS2=16 μeV\mu_{\mathrm{S1}}=\mu_{\mathrm{S2}}=16\,\mu\mathrm{eV}, μC=−72 μeV\mu_{\mathrm{C}}=-72\,\mu\mathrm{eV}, and μD=14 μeV\mu_{\mathrm{D}}=14\,\mu\mathrm{eV}.

Figure 15. Anti-clockwise cycle of thermal-AND gate: The anti-clockwise cycle operates for all input logic states. However, the output current arising solely from the anti-clockwise cycle is small in magnitude for all three input configurations. This is because the Fermi-level of the C lead is kept very low; as a result, it pumps electrons into the excited state of QDb\mathrm{QD}_{\mathrm{b}} at a low rate, ΓBDC\Gamma_{\mathrm{BD}}^{\mathrm{C}} (shown by the thin arrow). However, when this contribution is added to the clockwise cycle currents, the resulting total current produces the AND gate output logic values. The parameters for the AND gate are: μS1=μS2=16 μeV\mu_{\mathrm{S1}} = \mu_{\mathrm{S2}} = 16~\mu\mathrm{eV}, μC=−72 μeV\mu_{\mathrm{C}} = -72~\mu\mathrm{eV}, and μD=14 μeV\mu_{\mathrm{D}} = 14~\mu\mathrm{eV}.

∣C⟩|C\rangle to ∣D⟩|D\rangle is not facilitated by either of the source leads due to the absence of hot electrons. Nevertheless, the anti-clockwise cycle continues to operate in the opposite direction, resulting in a small amount of heat current flowing into the drain lead. However, the output current ∣JQD∣|J_{\mathrm{Q}}^{\mathrm{D}}| remains below the threshold value JQ0J_{\mathrm{Q}}^{0}, thus corresponds to the output logic-0.

  • For input logic (1,0) or (0,1), the clockwise transition cycle operates simultaneously with the anti-clockwise cycle [Fig. 15] due to the presence of one hot source lead, which facilitates the transition from ∣A⟩|A\rangle to ∣C⟩|C\rangle (in clockwise cycle). Despite the activation of the clockwise cycle [Fig. 7], due to the chosen position of the chemical potentials of the drain, control, and source leads, in order to keep the two thermal-diodes in reverse bias condition w.r.t the D lead and in forward bias condition w.r.t the C lead, the output heat current ∣JQD∣|J_{\mathrm{Q}}^{\mathrm{D}}| still remains below threshold JQ0J_{\mathrm{Q}}^{0}, corresponding to the output logic-0.

  • For input logic (1,1), the contribution of the clockwise transition cycle increases significantly due to the presence of both S leads [Fig. 14], as the transition from ∣A⟩|A\rangle to ∣C⟩|C\rangle (in the clockwise cycle) is facilitated by both leads. Consequently, the overall output current ∣JQD∣|J_{\mathrm{Q}}^{\mathrm{D}}| increases and exceeds the threshold value JQ1J_{\mathrm{Q}}^{1}, thereby corresponding to the output logic-1.

Thermal-NOR Gate

To construct the thermal-NOR gate, the thermal-NOT gate setup is combined with the thermal-OR gate circuit. Hence, QDa_a is coupled to an additional inverting lead I, along with S1 and S2. This configuration functions as a quantum thermal NOR gate where the input logics are governed by the temperatures of the S1 and S2-leads, whereas, like the thermal-NOT gate, the heat-current at the I-lead JQIJ_{\mathrm{Q}}^{\mathrm{I}} determines the output logic. In this setup, the steady-state rates Figure 16 follow [cf. Eq. (8)]

Figure 16. Diagrammatic representation of the clockwise transition cycle for the NOR gate.

ΓACD=ΓCDIS1S2=ΓDBD=ΓBAIS12≡Γ↻NOR(31)\Gamma_{\mathrm{AC}}^{\mathrm{D}}=\Gamma_{\mathrm{CD}}^{\mathrm{IS1S2}}=\Gamma_{\mathrm{DB}}^{\mathrm{D}}=\Gamma_{\mathrm{BA}}^{\mathrm{IS12}}\equiv\Gamma_{\circlearrowright}^{\mathrm{NOR}} \tag*{(31)}

where ΓijIS1S2≡ΓijI+ΓijS1+ΓijS2\Gamma_{ij}^{\mathrm{IS1S2}}\equiv\Gamma_{ij}^{\mathrm{I}}+\Gamma_{ij}^{\mathrm{S1}}+\Gamma_{ij}^{\mathrm{S2}}.

Now, just like the NOT gate, using the above relations in Eq. (14), the steady-state energy and particle currents associated with the I lead can be obtained as

JEI=εaΓABI+(εa+U)ΓCDI;JNI=ΓABI+ΓCDI.(32)J_{\mathrm{E}}^{\mathrm{I}}=\varepsilon_{\mathrm{a}}\Gamma_{\mathrm{AB}}^{\mathrm{I}}+(\varepsilon_{\mathrm{a}}+U)\Gamma_{\mathrm{CD}}^{\mathrm{I}};\qquad J_{\mathrm{N}}^{\mathrm{I}}=\Gamma_{\mathrm{AB}}^{\mathrm{I}}+\Gamma_{\mathrm{CD}}^{\mathrm{I}}. \tag*{(32)}

Therefore, the heat current JQIJ_{\mathrm{Q}}^{\mathrm{I}} with the invert lead reads,

JQI=JEI−μIJNI=(εa−μI)ΓABI+(εa+U−μI)ΓCDI=(εa−μI)(ΓCDI−ΓBAI)+UΓCDI.(33)J_{\mathrm{Q}}^{\mathrm{I}}=J_{\mathrm{E}}^{\mathrm{I}}-\mu_{\mathrm{I}}J_{\mathrm{N}}^{\mathrm{I}}=(\varepsilon_{\mathrm{a}}-\mu_{\mathrm{I}})\Gamma_{\mathrm{AB}}^{\mathrm{I}}+(\varepsilon_{\mathrm{a}}+U-\mu_{\mathrm{I}})\Gamma_{\mathrm{CD}}^{\mathrm{I}}=(\varepsilon_{\mathrm{a}}-\mu_{\mathrm{I}})(\Gamma_{\mathrm{CD}}^{\mathrm{I}}-\Gamma_{\mathrm{BA}}^{\mathrm{I}})+U\Gamma_{\mathrm{CD}}^{\mathrm{I}}. \tag*{(33)}

The above analytical expression gives the output heat-current JQIJ_{\mathrm{Q}}^{\mathrm{I}} and hence the output logic values. Now, the logical operations of the thermal-NOR gate can be explained by combining the thermal-NOT gate operations with the thermal-OR gate.

  • For input logic (0,0), the logical operation follows a pathway similar to that of the input logic-0 of the NOT gate. As illustrated in Figure 17, the transition from ∣C⟩\lvert C\rangle to ∣D⟩\lvert D\rangle is predominantly driven by the I-lead due to

Figure 17. Thermal-NOR gate is implemented by applying a NOT operation to an OR gate, achieved by introducing an additional I lead coupled to QDₐ. In contrast to the (0, 0) case of the OR gate, the presence of the hot I-lead drives the cycle. Since the output current is measured at the I-lead, this configuration yields output logic-1. When one of the inputs is 1, the heat-flow rate from the I-lead is suppressed, and the measured current drops below threshold, resulting in output logic-0. The parameters for the NOR gate are: μS1=μS2=−95 μeV\mu_{\mathrm{S1}}=\mu_{\mathrm{S2}}=-95~\mu\mathrm{eV}, μI=−90 μeV\mu_{\mathrm{I}}=-90~\mu\mathrm{eV}, and μD=21 μeV\mu_{\mathrm{D}}=21~\mu\mathrm{eV}.

the absence of hot electrons from both cold S leads, whereas the de-excitation process from ∣B⟩\lvert B\rangle to ∣A⟩\lvert A\rangle is shared between the I and both S leads, where cold source leads take the advantage. Furthermore, the parameters of thermal-NOR gate satisfy μI<0\mu_I < 0 and εa,U>0\varepsilon_a, U > 0. Substituting these conditions into Eq. (33), it follows that the output current JQI>0J_Q^I > 0, but also exceeds the threshold value JQ0J_Q^0, thereby corresponding to the output logic-1.

  • For input logic (1,0) or (0,1), one of the source leads provides hot electrons to facilitate the transition from ∣C⟩\lvert C\rangle to ∣D⟩\lvert D\rangle along with the I-lead, thereby effectively reducing the dominant contribution of the I-lead to this transition. On the other hand, the de-excitation rate corresponding to the I-lead ΓBAI\Gamma_{BA}^I increases, since the contribution from the hot S-lead to the de-excitation process decreases [Fig. 17]. Consequently, following Eq. (33), above conditions drive the output current JQIJ_Q^I below the threshold value JQ0J_Q^0, corresponding to output logic-0.

  • For input logic (1,1), the participation of the I-lead in the transition from ∣C⟩\lvert C\rangle to ∣D⟩\lvert D\rangle is significantly reduced because both hot source leads actively facilitate this transition, whereas ΓBAI\Gamma_{BA}^I becomes maximal due to the absence of any cold source lead [Fig. 17]. Consequently, as dictated by Eq. (33), the output current JQIJ_Q^I decreases substantially and remains well below the threshold value JQ0J_Q^0, thereby corresponding to the output logic-0.

Thermal-NAND Gate

Following a similar idea, a thermal-NAND gate can be constructed by combining a thermal-NOT gate with the thermal-AND gate circuit. In this configuration, QDa_a is coupled to an additional inverting lead I, along with the two input reservoirs S1 and S2, whereas QDb_b is coupled to D and C-lead, as in the thermal-AND gate.

Figure 18. Thermal-NOR gate anticlockwise cycle: Since no hot lead is attached to QDb_b, during the second step of each input logic operation, the cold D lead cannot inject an electron into the excited state of QDb_b. Thus, none of the anti-clockwise cycles can be completed, and similar to the buffer, NOT, and OR gates, the output current arising from the anti-clockwise cycles in the NOR gate is ultimately negligible. The parameters for the NOR gate are: μS1=μS2=−95 μeV\mu_{S1} = \mu_{S2} = -95\,\mu\mathrm{eV}, μI=−90 μeV\mu_I = -90\,\mu\mathrm{eV}, and μD=21 μeV\mu_D = 21\,\mu\mathrm{eV}.

As before, the input logic states are controlled by the temperatures of the leads S1 and S2. Meanwhile, similar to the NOT gate, the output logic is determined by the magnitude of the heat current associated with the inverting lead, i.e., JQIJ_Q^I. Under these conditions, the steady-state transition rates [Figure 19] satisfy relations analogous to those given in Eq. (8):

Figure 19. Diagrammatic representation for clockwise and anti-clockwise transition cycles for NAND gate.

ΓACDC=ΓCDIS1S2=ΓDBDC=ΓBAIS1S2≡Γ↻NAND ;ΓABIS1S2=ΓBDDC=ΓDCIS1S2=ΓCADC≡Γ↺NAND(34)\Gamma_{\mathrm{AC}}^{\mathrm{DC}}=\Gamma_{\mathrm{CD}}^{\mathrm{IS1S2}}=\Gamma_{\mathrm{DB}}^{\mathrm{DC}}=\Gamma_{\mathrm{BA}}^{\mathrm{IS1S2}}\equiv\Gamma_{\circlearrowright}^{\mathrm{NAND}}\,;\qquad\Gamma_{\mathrm{AB}}^{\mathrm{IS1S2}}=\Gamma_{\mathrm{BD}}^{\mathrm{DC}}=\Gamma_{\mathrm{DC}}^{\mathrm{IS1S2}}=\Gamma_{\mathrm{CA}}^{\mathrm{DC}}\equiv\Gamma_{\circlearrowleft}^{\mathrm{NAND}} \tag*{(34)}

where ΓijIS1S2=ΓijI+ΓijS1+ΓijS2\Gamma_{ij}^{\mathrm{IS1S2}}=\Gamma_{ij}^{\mathrm{I}}+\Gamma_{ij}^{\mathrm{S1}}+\Gamma_{ij}^{\mathrm{S2}} and ΓijDC=ΓijD+ΓijC\Gamma_{ij}^{\mathrm{DC}}=\Gamma_{ij}^{\mathrm{D}}+\Gamma_{ij}^{\mathrm{C}}, with Γ↺NAND=−Γ↻NAND\Gamma_{\circlearrowleft}^{\mathrm{NAND}}=-\Gamma_{\circlearrowright}^{\mathrm{NAND}}. Substituting these relations into Eq. (14), the steady-state energy and particle currents associated with the inverting lead are obtained as

JEI=εaΓABI+(εa+U)ΓCDI ;JNI=ΓABI+ΓCDI.(35)J_{\mathrm{E}}^{\mathrm{I}}=\varepsilon_a\Gamma_{\mathrm{AB}}^{\mathrm{I}}+(\varepsilon_a+U)\Gamma_{\mathrm{CD}}^{\mathrm{I}}\,;\qquad J_{\mathrm{N}}^{\mathrm{I}}=\Gamma_{\mathrm{AB}}^{\mathrm{I}}+\Gamma_{\mathrm{CD}}^{\mathrm{I}}. \tag*{(35)}

Accordingly, the heat current associated with the inverting lead can be written as

JQI=JEI−μIJNI=(εa−μI)ΓABI+(εa+U−μI)ΓCDI=(εa−μI)(ΓABI−ΓDCI)−UΓDCI=(εa−μI)(ΓCDI−ΓBAI)+UΓCDI.(36)\begin{aligned} J_{\mathrm{Q}}^{\mathrm{I}}&=J_{\mathrm{E}}^{\mathrm{I}}-\mu_{\mathrm{I}}J_{\mathrm{N}}^{\mathrm{I}}=(\varepsilon_a-\mu_{\mathrm{I}})\Gamma_{\mathrm{AB}}^{\mathrm{I}}+(\varepsilon_a+U-\mu_{\mathrm{I}})\Gamma_{\mathrm{CD}}^{\mathrm{I}}\\ &=(\varepsilon_a-\mu_{\mathrm{I}})(\Gamma_{\mathrm{AB}}^{\mathrm{I}}-\Gamma_{\mathrm{DC}}^{\mathrm{I}})-U\Gamma_{\mathrm{DC}}^{\mathrm{I}}\\ &=(\varepsilon_a-\mu_{\mathrm{I}})(\Gamma_{\mathrm{CD}}^{\mathrm{I}}-\Gamma_{\mathrm{BA}}^{\mathrm{I}})+U\Gamma_{\mathrm{CD}}^{\mathrm{I}}. \tag*{(36)} \end{aligned}

The above analytical expression of JQIJ_{\mathrm{Q}}^{\mathrm{I}} determines the thermal logic values. Similar to the NOR gate, the logical operation of the NAND gate can be understood as a combination of the NOT and AND gate operations. Like in the AND gate configuration, both clockwise and anti-clockwise transition cycles may occur due to the coupling of QDbQ_{\mathrm{D}b} with the hot C lead. Therefore, as indicated by Eq. (36), the relevant transitions are those between ∣A⟩|A\rangle and ∣B⟩|B\rangle, and between ∣C⟩|C\rangle and ∣D⟩|D\rangle. In the clockwise transition cycle [See Figure 20], the I-lead participates in both the excitation

Figure 20. A thermal-NAND gate is realized by applying a NOT operation to an AND gate. Like the AND gate, the NAND gate cycle can also run in both directions. Only the ↻\circlearrowright cycle is shown here; cf. Figure 21 for the ↺\circlearrowleft cycle. Since the output is measured at the I-lead, the presence of any cold input (logic-0) ensures sufficient heat current at I (due to both ↻\circlearrowright and ↺\circlearrowleft cycles), resulting in output logic-1. Only when both inputs are 1, S and I equally contribute in heat-flow, thus heat-current at I drops significantly, resulting in logic-0. Here, arrow widths indicate relative heat-flow strengths. The parameters for the NAND gate are: μS1=μS2=16 μeV\mu_{\mathrm{S1}}=\mu_{\mathrm{S2}}=16~\mu\mathrm{eV}, μC=−16 μeV\mu_{\mathrm{C}}=-16~\mu\mathrm{eV}, μI=16 μeV\mu_{\mathrm{I}}=16~\mu\mathrm{eV}, and μD=9 μeV\mu_{\mathrm{D}}=9~\mu\mathrm{eV}.

process from ∣C⟩\lvert C\rangle to ∣D⟩\lvert D\rangle and the de-excitation process from ∣B⟩\lvert B\rangle to ∣A⟩\lvert A\rangle, resulting in a finite heat current flowing from the I-lead. The magnitude of this current depends on the input logic states; it gradually decreases as the number of hot input-sources increases.

On the other hand, in the anti-clockwise transition cycle Figure 21, a finite heat current is also associated with the I-lead, as it contributes to both the excitation from ∣A⟩\lvert A\rangle to ∣B⟩\lvert B\rangle and the de-excitation from ∣D⟩\lvert D\rangle to ∣C⟩\lvert C\rangle. However, in this case, the heat current is comparatively small and insensitive to changes in the input logic states.

Figure 21. Thermal-NAND gate anticlockwise cycle: Unlike the AND gate, in this case, the Fermi level of the C lead is kept significantly higher than that in the AND gate configuration. As a result, the C lead can supply electrons into QDb\mathrm{QD}_{b} with a nonzero transition rate, ΓBDC\Gamma_{\mathrm{BD}}^{\mathrm{C}}, resulting in a finite output current measured at the I lead. When this current is combined with the clockwise cycle current, it produces a large output current at the I lead. The parameters for the NAND gate are: μS1=μS2=16 μeV\mu_{\mathrm{S1}} = \mu_{\mathrm{S2}} = 16\,\mu\mathrm{eV}, μC=−16 μeV\mu_{\mathrm{C}} = -16\,\mu\mathrm{eV}, μI=16 μeV\mu_{\mathrm{I}} = 16\,\mu\mathrm{eV}, and μD=9 μeV\mu_{\mathrm{D}} = 9\,\mu\mathrm{eV}.

  • For input logic (0,0), both the excitation process from ∣C⟩\lvert C\rangle to ∣D⟩\lvert D\rangle and the de-excitation process from ∣B⟩\lvert B\rangle to ∣A⟩\lvert A\rangle are predominantly governed by the I-lead due to the presence of both cold source leads [See Figure 20]. Consequently, the clockwise transition cycle generates a substantial output current. In addition to the clockwise cycle [Figure 20], the anti-clockwise cycle [Figure 21] also operates simultaneously and contributes to the overall output current. Owing to the combined contribution of both cycles, particularly the dominant clockwise cycle, a significantly large output heat current JQIJ_{\mathrm{Q}}^{\mathrm{I}} flows from the I-lead. This current remains well above the threshold value JQ1J_{\mathrm{Q}}^{1}, thereby corresponding to the output logic-1.

  • For input logic (1,0) or (0,1), the presence of one hot source lead reduces the contribution of the I-lead in the transitions from ∣C⟩\lvert C\rangle to ∣D⟩\lvert D\rangle and from ∣B⟩\lvert B\rangle to ∣A⟩\lvert A\rangle. Consequently, the heat current associated with the I-lead in the clockwise transition cycle [Figure 20] decreases, although it remains appreciable since one of the source leads still acts as a cold bath. Alongside the clockwise cycle, the anti-clockwise cycle [Figure 21] also persists; however, as discussed earlier, its contribution is not significantly affected by changes in the input logic states. Therefore, due to the combined contributions from both cycles, the output current JQIJ_{\mathrm{Q}}^{\mathrm{I}} still remains above the threshold value JQ1J_{\mathrm{Q}}^{1}, thereby corresponding to the output logic-1.

  • For input logic (1,1), due to the absence of any cold source lead, the contribution of the I-lead to the transitions from ∣C⟩\lvert C\rangle to ∣D⟩\lvert D\rangle and from ∣B⟩\lvert B\rangle to ∣A⟩\lvert A\rangle decreases significantly [Figure 20]. Consequently, the output current associated with the clockwise transition cycle is substantially reduced. Although the anti-clockwise cycle [Figure 21] still contributes to the output current to some extent, the dominant contribution originates from the clockwise cycle. Therefore, as this contribution decreases markedly, the overall output current falls below the lower threshold value JQ0J_{\mathrm{Q}}^{0}, thereby corresponding to the output logic-0.

References

  1. [1]R. Maurand and X. Jehl, Transistor qubits heat up, Nature Electronics 5, 131 (2022).DOI
  2. [2]M. A. Aamir, P. Jamet Suria, J. A. Marín Guzmán, C. Castillo-Moreno, J. M. Epstein, N. Yunger Halpern, and S. Gasparinetti, Thermally driven quantum refrigerator autonomously resets a superconducting qubit, Nature Physics 21, 318 (2025).DOI
  3. [3]M. S. Blok and G. T. Landi, Quantum thermodynamics for quantum computing, Nature Physics 21, 187 (2025).DOI
  4. [4]B. Karimi, G. O. Steffensen, A. P. Higginbotham, C. M. Marcus, A. Levy Yeyati, and J. P. Pekola, Bolometric detection of josephson radiation, Nature Nanotechnology 19, 1613 (2024).arxiv.org/abs/2402.09314
  5. [5]S. Simbierowicz, M. Borrelli, V. Monarkha, V. Nuutinen, and R. E. Lake, Inherent thermal-noise problem in addressing qubits, PRX Quantum 5, 030302 (2024).DOI
  6. [6]V. Champain, G. Boschetto, H. Niebojewski, B. Bertrand, L. Mauro, M. Bassi, V. Schmitt, X. Jehl, S. Zihlmann, R. Maurand, Y. M. Niquet, C. B. Winkelmann, S. D. Franceschi, B. Martinez, and B. Brun, A heat-resilient hole spin qubit in silicon (2025), arXiv:2509.15823.
  7. [7]J. Roßnagel, S. T. Dawkins, K. N. Tolazzi, O. Abah, E. Lutz, F. Schmidt-Kaler, and K. Singer, A single-atom heat engine, Science 352, 325 (2016).
  8. [8]H. Thierschmann, R. Sánchez, B. Sothmann, F. Arnold, C. Heyn, W. Hansen, H. Buhmann, and L. W. Molenkamp, Three-terminal energy harvester with coupled quantum dots, Nature Nanotechnology 10, 854 (2015).DOI
  9. [9]S. Jezouin, F. D. Parmentier, A. Anthore, U. Gennser, A. Cavanna, Y. Jin, and F. Pierre, Quantum limit of heat flow across a single electronic channel, Science 342, 601 (2013).DOI
  10. [10]M. Banerjee, M. Heiblum, A. Rosenblatt, Y. Oreg, D. E. Feldman, A. Stern, and V. Umansky, Observed quantization of anyonic heat flow, Nature 545, 75 (2017).DOI
  11. [11]B. Dutta, V. Umansky, M. Banerjee, and M. Heiblum, Isolated ballistic non-abelian interface channel, Science 377, 1198 (2022).DOI
  12. [12]N. M. Myers, O. Abah, and S. Deffner, Quantum thermodynamic devices: From theoretical proposals to experimental reality, AVS Quantum Science 4, 027101 (2022), and the references therein.DOI
  13. [13]K. Joulin, J. Drevillon, Y. Ezzahri, and J. Ordonez-Miranda, Quantum thermal transistor, Phys. Rev. Lett. 116, 200601 (2016).DOI
  14. [14]B. Dutta, D. Majidi, A. García Corral, P. A. Erdman, S. Florens, T. A. Costi, H. Courtois, and C. B. Winkelmann, Direct probe of the seebeck coefficient in a kondocorrelated single-quantum-dot transistor, Nano Letters 19, 506 (2019).DOI
  15. [15]M. L. Perrin, E. Burzurí, and H. S. J. van der Zant, Single-molecule transistors, Chem. Soc. Rev. 44, 902 (2015).DOI
  16. [16]N. Gupt, S. Bhattacharyya, B. Das, S. Datta, V. Mukherjee, and A. Ghosh, Floquet quantum thermal transistor, Phys. Rev. E 106, 024110 (2022).DOI
  17. [17]R. T. Wijesekara, S. D. Gunapala, and M. Premaratne, Darlington pair of quantum thermal transistors, Phys. Rev. B 104, 045405 (2021).DOI
  18. [18]M. L. Perrin, E. Galán, R. Eelkema, J. M. Thijssen, F. Grozema, and H. S. J. van der Zant, A gate-tunable single-molecule diode, Nanoscale 8, 8919 (2016).DOI
  19. [19]S. Ghosh, N. Gupt, and A. Ghosh, Universal behavior of the coulomb-coupled fermionic thermal diode, Entropy 24 (2022).DOI
  20. [20]W. Khan, P. P. Potts, S. Lehmann, C. Thelander, K. A. Dick, P. Samuelsson, and V. F. Maisi, Efficient and continuous microwave photoresponse in hybrid cavity-semiconductor nanowire double quantum dot diodes, Nature Communications 12, 5130 (2021).DOI
  21. [21]R. Upadhyay, D. S. Golubov, Y.-C. Chang, G. Thomas, A. Guthrie, J. T. Peltonen, and J. P. Pekola, Microwave quantum diode, Nature Communications 15, Article number: 44908 (2024).DOI
  22. [22]M. Moskalets, P. Samuelsson, and M. Büttiker, Quantized dynamics of a coherent capacitor, Phys. Rev. Lett. 100, 086601 (2008).DOI
  23. [23]F. Campaioli, S. Gherardini, J. Q. Quach, M. Polini, and G. M. Andolina, Colloquium: Quantum batteries, Rev. Mod. Phys. 96, 031001 (2024).DOI
  24. [24]D. Ferraro, F. Cavalcav, M. G. Genoni, G. Benenti, and M. Sassetti, Opportunities and challenges of quantum batteries, Nature Reviews Physics 8, 115 (2026).DOI
  25. [25]R. Kosloff and A. Levy, Quantum heat engines and refrigerators: Continuous devices, Annu. Rev. Phys. Chem. 65, 365 (2014).DOI
  26. [26]A. Ghosh, C. L. Latune, L. Davidovich, and G. Kurizki, Catalysis of heat-to-work conversion in quantum machines, Proceedings of the National Academy of Sciences 114, 12156 (2017).DOI
  27. [27]A. Ghosh, D. Gelbwaser-Klimovsky, W. Niedenzu, A. I. Lvovsky, I. Mazets, M. O. Scully, and G. Kurizki, Two-level masers as heat-to-work converters, Proceedings of the National Academy of Sciences 115, 9941 (2018).DOI
  28. [28]F. Binder, L. A. Correa, C. Gogolin, J. Anders, and G. Adesso, Thermodynamics in the quantum regime: Fundamental aspects and new directions (Springer, 2019) and the references therein.
  29. [29]A. Maity, P. Chaki, A. Ghoshal, and U. Sen, Quantum heat transformers, New Journal of Physics 28, 014501 (2025).DOI
  30. [30]G. Kurizki and A. Kofman, Thermodynamics and Control of Open Quantum Systems (Cambridge University Press, 2022).DOI
  31. [31]F. Giazotto, T. T. Heikkilä, A. Luukanen, A. M. Savin, and J. P. Pekola, Opportunities for mesoscopics in thermometry and refrigeration: Physics and applications, Rev. Mod. Phys. 78, 217 (2006).DOI
  32. [32]J. P. Pekola, K. P. Hirvi, J. P. Kauppinen, and M. A. Paalanen, Thermometry by arrays of tunnel junctions, Phys. Rev. Lett. 73, 2903 (1994).pubmed.ncbi.nlm.nih.gov/10057225
  33. [33]M. Nahum and J. M. Martinis, Ultrasensitive hot-electron microbolometer, Applied Physics Letters 63, 3075 (1993).DOI
  34. [34]M. Meschke, J. T. Peltonen, H. Courtois, and J. P. Pekola, Calorimetric readout of a superconducting proximity-effect thermometer, Journal of Low Temperature Physics 154, 190 (2009).DOI
  35. [35]D. Halbertal, J. Cuppens, M. Ben Shalom, L. Embon, N. Shadmi, Y. Anahory, H. R. Naren, J. Sarkar, A. Uri, Y. Ronen, Y. Myasoedov, L. S. Levitov, E. Joselevich, A. K. Geim, and E. Zeldov, Nanoscale thermal imaging of dissipation in quantum systems, Nature 539, 407 (2016).DOI
  36. [36]B. Karimi, F. Brange, P. Samuelsson, and J. P. Pekola, Reaching the ultimate energy resolution of a quantum detector, Nature Communications 11, 367 (2020).DOI
  37. [37]J. P. Pekola and B. Karimi, Colloquium: Quantum heat transport in condensed matter systems, Rev. Mod. Phys. 93, 041001 (2021).DOI
  38. [38]D. Majidi, J. P. Bergfield, V. Maisi, J. Höfer, H. Courtois, and C. B. Winkelmann, Heat transport at the nanoscale and ultralow temperatures—implications for quantum technologies, Applied Physics Letters 124, 140504 (2024).DOI
  39. [39]B. Dutta, J. T. Peltonen, D. S. Antonenko, M. Meschke, M. A. Skvortsov, B. Kubala, J. König, C. B. Winkelmann, H. Courtois, and J. P. Pekola, Thermal conductance of a single-electron transistor, Phys. Rev. Lett. 119, 077701 (2017).DOI
  40. [40]L. Cui, W. Jeong, S. Hur, M. Matt, J. C. Klöckner, F. Pauly, P. Nielaba, J. C. Cuevas, E. Meyhofer, and P. Reddy, Quantized thermal transport in single-atom junctions, Science 355, 1192 (2017).DOI
  41. [41]N. Mosso, U. Drechsler, F. Menges, P. Nirmalraj, S. Karg, H. Riel, and B. Gotsmann, Heat transport through atomic contacts, Nature Nanotechnology 12, 430 (2017).DOI
  42. [42]M. Josefsson, A. Svilans, A. M. Burke, E. A. Hoffmann, S. Fahlvik Svensson, C. Thelander, M. Leijnse, and H. Linke, A quantum-dot heat engine operating close to the thermodynamic efficiency limits, Nature Nanotechnology 13, 920 (2018).DOI
  43. [43]S. Volosheniuk, R. Conte, E. Pyurbeeva, T. Baum, M. Vilas-Varela, S. Fernández, D. Peña, H. S. J. van der Zant, and P. Gehring, A single-molecule quantum heat engine, Nano Letters 26, 984 (2026).arxiv.org/abs/2508.17036
  44. [44]B. Dutta, D. Majidi, N. W. Talarico, N. Lo Gullo, H. Courtois, and C. B. Winkelmann, Single-quantum-dot heat valve, Phys. Rev. Lett. 125, 237701 (2020).DOI
  45. [45]O. Maillet, D. Subero, J. T. Peltonen, D. S. Golubev, and J. P. Pekola, Electric field control of radiative heat transfer in a superconducting circuit, Nature Communications 11, 4326 (2020).DOI
  46. [46]A. Ronzani, B. Karimi, J. Senior, Y.-C. Chang, J. T. Peltonen, C. Chen, and J. P. Pekola, Tunable photonic heat transport in a quantum heat valve, Nature Physics 14, 991 (2018).DOI
  47. [47]S. Campbell et al., Roadmap on quantum thermodynamics, Quantum Science and Technology 11, 012501 (2026), and the references therein.DOI
  48. [48]D. Tiwari, S. Bhattacharya, and S. Banerjee, Quantum thermal analogs of electric circuits: A universal approach, Phys. Rev. Lett. 135, 020404 (2025).DOI
  49. [49]L. Wang, Z. Wang, C. Wang, and J. Ren, Cycle flux ranking of network analysis in quantum thermal devices, Phys. Rev. Lett. 128, 067701 (2022).DOI
  50. [50]L. Tesser, B. Bhandari, P. A. Erdman, E. Paladino, R. Fazio, and F. Taddei, Heat rectification through single and coupled quantum dots, New Journal of Physics 24, 035001 (2022).
  51. [51]N. Gupt, S. Ghosh, and A. Ghosh, Graph theoretic analysis of three-terminal quantum dot thermocouples: Onsager relations and spin-thermoelectric effects, Phys. Rev. B 109, 125124 (2024).DOI
  52. [52]S. Ghosh, N. Gupt, and A. Ghosh, Inverse current in coupled transport: A quantum thermodynamic model, Phys. Rev. Res. 8, 023166 (2026).DOI
  53. [53]L. Wang and B. Li, Thermal logic gates: Computation with phonons, Phys. Rev. Lett. 99, 177208 (2007).DOI
  54. [54]F. Paolucci, G. Marchegiani, E. Strambini, and F. Giazotto, Phase-tunable thermal logic: Computation with heat, Phys. Rev. Appl. 10, 024003 (2018).DOI
  55. [55]T. Ruokola and T. Ojanen, Single-electron heat diode: Asymmetric heat transport between electronic reservoirs through coulomb islands, Phys. Rev. B 83, 241404 (2011).DOI
  56. [56]J. V. Koski, A. Kutvonen, I. M. Khaymovich, T. Ala-Nissila, and J. P. Pekola, On-chip maxwell’s demon as an information-powered refrigerator, Phys. Rev. Lett. 115, 260602 (2015).DOI
  57. [57]H. Thierschmann, F. Arnold, M. Mitttermüller, L. Maier, C. Heyn, W. Hansen, H. Buhmann, and L. W. Molenkamp, Thermal gating of charge currents with coulomb coupled quantum dots, New Journal of Physics 17, 113003 (2015).DOI
  58. [58]Y. Zhang, X. Zhang, Z. Ye, G. Lin, and J. Chen, Three-terminal quantum-dot thermal management devices, Applied Physics Letters 110, 153501 (2017).DOI
  59. [59]R. S. Whitney, R. Sánchez, and J. Splettstoesser, Quantum thermodynamics of nanoscale thermoelectrics and electronic devices, in Thermodynamics in the Quantum Regime: Fundamental Aspects and New Directions, edited by F. Binder, L. A. Correa, C. Gogolin, J. Anders, and G. Adesso (Springer International Publishing, Cham, 2018) pp. 175–206.
  60. [60]R. Sánchez and M. Büttiker, Optimal energy quanta to current conversion, Phys. Rev. B 83, 085428 (2011).DOI
  61. [61]E. Pyurbeeva and R. Kosloff, Quantum dot thermal machines—a guide to engineering, Entropy 28, 10.3390/e28010002 (2026).arxiv.org/abs/2511.04324
  62. [62]A. Ghosh, S. S. Sinha, and D. S. Ray, Fermionic oscillator in a fermionic bath, Phys. Rev. E 86, 011138 (2012).pubmed.ncbi.nlm.nih.gov/23005399
  63. [63]N. Gupt, S. Bhattacharya, and A. Ghosh, Statistical generalization of regenerative bosonic and fermionic stir-ling cycles, Phys. Rev. E 104, 054130 (2021).DOI
  64. [64]G. G. Damas, R. J. de Assis, and N. G. de Almeida, Cooling with fermionic thermal reservoirs, Phys. Rev. E 107, 034128 (2023).DOI
  65. [65]H.-P. Breuer and F. Petruccione, The Theory of Open Quantum Systems (Oxford University Press, 2007).
  66. [66]P. Strasberg, Quantum Stochastic Thermodynamics: Foundations and Selected Applications (Oxford University Press, 2022).DOI
  67. [67]J. Millman and C. Halkias, Integrated Electronics: Analog and Digital Circuits and Systems, Electrical Engineering Series (McGraw-Hill, 1972).
  68. [68]P. Dubos, H. Courtois, B. Pannetier, F. K. Wilhelm, A. D. Zaikin, and G. Schön, Josephson critical current in a long mesoscopic s-n-s junction, Phys. Rev. B 63, 064502 (2001).DOI
  69. [69]M. Esposito, K. Lindenberg, and C. V. den Broeck, Entropy production as correlation between system and reservoir, New Journal of Physics 12, 013013 (2010).DOI
  70. [70]G. T. Landi and M. Paternostro, Irreversible entropy production: From classical to quantum, Rev. Mod. Phys. 93, 035008 (2021).DOI

Paper details

vlnc:W_uwjnc8zc
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents