Barends five-qubit (Metal 52)

Five Xmon qubits sit in a line, close enough that neighbours couple directly through their facing cross arms. Each has its own λ/4 readout meander on a shared feedline (whose middle section acts as a Purcell filter), an XY drive line and a Z flux line. It is the Qiskit Metal “52” example (Apache-2.0, © IBM and contributors), a layout inspired by Barends et al. (Nature 508, 500, 2014), not the fabricated mask. The Quantcore rebuild is distributed under GPL-3.0.

Build this one to learn direct qubit-qubit coupling without bus resonators, multiplexed readout with staggered resonator lengths, and a full set of drive and flux lines on a 7.5 mm die.

The whole m52 chip as the v2 notebook builds it

At a glance

Notebook

examples/designs/metal/m52_barends_5q.ipynb (an older one-cell version is kept as m52_barends_5q.v1.ipynb)

Source

Qiskit Metal 52-Barends_5Qubit_Xmon_Processor.ipynb, © IBM and contributors, Apache-2.0

Chip

7.5 × 7.5 mm Quantcore die (the source’s 6 × 6 mm area centred in it), CPW 10 / 6 µm

On the chip

5 flux-tunable Xmons (450 µm pitch) · 5 λ/4 readouts (4.41 to 4.57 mm) · 5 coupled-line tees · 2 Purcell capacitors · 5 XY and 5 Z lines · 12 bond pads

Builds today

yes (layout built on 2026-11-19 in 13 s)

How the notebook builds it

The layout segments of m52, one picture each
  1. Die. The 7.5 mm die with the QCT frame (dicing marks, alignment grids, chip boundary, device ID).

  2. Bond pads. Twelve launchpads, three per edge, slid to the 100 µm pad-edge rule (the picture zooms on one pad).

  3. Feedline. Five coupled-line tees, two Purcell series capacitors and the feedline between the input and output pads.

  4. Qubits. Five Metal crossmons with Quantcore SQUIDs, 450 µm apart.

  5. Readout resonators. Five λ/4 meanders from each qubit’s claw to its tee.

  6. Coupling. The four direct neighbour couplings, declared in the netlist (the picture zooms on the qubits).

  7. Control lines. An XY line ending open beside each cross and a Z flux line for each qubit, to the edge pads.

  8. Whole chip. Sim groups for the FEM jobs (the qubit chain and the readout), the content centred in the die, the design saved.

LOM estimates

Numbers are from a run of this notebook on 2026-11-19 (Setup, Design, Check and the LOM cells, 38 s): QuCat normal modes with QuTiP dressed values.

group

qubit

f_q (GHz)

α (MHz)

resonator

f_r (GHz)

κ (MHz)

χ (MHz)

Q0+R0

Q0

6.357

-277

CLT0

6.275

0

28

Q1+R1

Q1

6.352

-272

CLT1

6.23

0

23.3

Q2+R2

Q2

6.35

-266

CLT2

6.183

0

23.9

Q3+R3

Q3

6.348

-273

CLT3

6.136

0

30.2

Q4+R4

Q4

6.348

-212

CLT4

6.089

0

-31.3

group

pair

ZZ (kHz)

neighbours

hybridised

chain

Q1-Q2

–

yes

yes

chain

Q0-Q2

–

no

yes

chain

Q2-Q4

–

no

yes

chain

Q2-Q3

–

yes

yes

chain

Q0-Q1

–

yes

yes

chain

Q1-Q4

–

no

yes

chain

Q1-Q3

–

no

yes

chain

Q0-Q4

–

no

yes

chain

Q0-Q3

–

no

yes

chain

Q3-Q4

–

yes

yes

LOM plot of m52_barends_5q

These are lumped-model estimates from component formulas and calibration, not FEM results. The LOM page says how far to trust them.

Build spec

The rest of this page is the detailed spec the notebook was built from: every component, port, route and length.

The metal: references in this spec build as written, through the metal: wrapper (ports are the Metal pin names, parameters the Metal options). The Quantcore and KQC substitutions are optional.

Source: Quantum Metal docs/circuit-examples/F.Small-quantum-chips/52-Barends_5Qubit_Xmon_Processor.ipynb. Licence Apache-2.0. The notebook adapts the QDW 2026 capstone solution by M. C. Sarihan, after Barends et al., Nature 508, 500 (2014). The notebook calls it “a physical-layout example inspired by the published device, not the fabricated mask”. The chip has five TransmonCrossFL Xmons in a 1-D chain at 450 µm pitch. Their ground pockets overlap, so neighbours couple directly through the arm capacitance and there are no bus resonators. Each qubit has a λ/4 meander readout shorted at a CoupledLineTee on one feedline. The feedline section between two CapNInterdigital series caps acts as a common Purcell filter. Each qubit also has an XY drive line (open end) and a Z flux line, each going to its own edge pad. Chip 6000 × 6000 µm, centred, so Designer = Metal + (3000, 3000). CPW 10/6 µm except the Z-line pin (5 µm); all routes in the source use 10/6. Metal layer 1. The junctions are left out of the GDS (no junction file). A first build uses these substitutions: Xmon with flux_port, hanger for CoupledLineTee, KQC FingerCapacitorSquare for CapNInterdigital, BondPad, and free ends for the XY lines.

Components

instance

catalog ref

key params (µm)

at (x, y)

rot

notes

Q0 … Q4

Xmon (metal:TransmonCrossFL)

Metal cross_width 30, cross_length 190 (arm), cross_gap 50 → Xmon arm_width=30, cross_length=380, gap=50, claws=["n"], claw_length=40, claw_base=10, claw_finger_width=10, claw_spacing=6, ground_sliver=5, flux_port=True

(2100 + 450 i, 3000)

0

readout claw north (Metal connector_location 90); flux T-line south (t_top 15, t_width 5, t_gap 3, t_inductive_gap 3)

CLT0 … CLT4

hanger on FL_P (metal:CoupledLineTee)

coupling_length 200, coupling_space 4, down_length 100, open_termination False (short)

centre (2100 + 450 i, 4500)

0

second line below the feed; leg at x + 100

XY0 … XY4

free end (metal:OpenToGround)

width 10, gap 6, termination_gap 6

(2000 + 450 i, 2820)

—

open drive end SW of each cross

CAP_IN

KQC FingerCapacitorSquare (metal:CapNInterdigital)

Metal: 5 fingers × 20 µm, cap_width 10, cap_gap 6, cap_distance 50

(527, 4500)

0

Metal span 454–600; a faces W, b faces E

CAP_OUT

KQC FingerCapacitorSquare

same

(2073, 5100)

0

Metal span 2000–2146

LP_W1, W2, W3

BondPad(pad_width=80, pad_height=80, pad_gap=58, taper_length=122)

Metal default pad

(225, 1500), (225, 3000), (225, 4500)

270

at = Metal tie pin

LP_E1, E2, E3

BondPad

same

(5775, 1500), (5775, 3000), (5775, 4500)

90

LP_S1, S2, S3

BondPad

same

(1500, 225), (3000, 225), (4500, 225)

0

LP_N1, N2, N3

BondPad

same

(1500, 5775), (3000, 5775), (4500, 5775)

180

Routes

route

from → to

type, length

waypoints / hanger

notes

FL_IN

LP_W3.a → CAP_IN.a

cpw ≈ 265

—

Metal Feed_in_lp

FL_P

CAP_IN.b → CAP_OUT.a

cpw ≈ 10 030

nodes [(4700, 4500), (4700, 4800), (600, 4800), (600, 5100)]

Purcell section; straight y = 4500 run carries the five hangers

FL_OUT

CAP_OUT.b → LP_N1.a

cpw ≈ 1695, r = 20

nodes [(2346, 5100), (2346, 5675), (1500, 5675)]

Metal lead 200/100

R0 … R4

Qi.claw_n → FL_P.hanger(s=1536.5+450i, length=200, side="right", ground_gap=4, direction="backward", end="short")

meander 4710 + 40 i

—

Metal total_length 4.41–4.57 mm + 300 µm CLT (down leg + coupler); spacing 50, fillet 20

XY0

LP_W2.a → free_end((2000, 2820), dir=-90, term=6)

cpw 2329, r = 20

[(275, 3000), (275, 2620), (2000, 2620)]

the XY rows below are Metal RouteFramed paths, reversed

Z0

LP_W1.a → Q0.flux

cpw 3088

[(2100, 1500)]

XY1

LP_S1.a → free_end((2450, 2820), -90, 6)

cpw 3511

[(1500, 620), (2400, 620), (2400, 2620), (2450, 2620)]

Z1

LP_S2.a → Q1.flux

cpw 2912

[(3000, 622), (2650, 622), (2650, 2622), (2550, 2622)]

XY2

LP_S3.a → free_end((2900, 2820), -90, 6)

cpw 4261

[(4500, 820), (2850, 820), (2850, 2620), (2900, 2620)]

Z2

LP_E1.a → Q2.flux

cpw 4911

[(5600, 1500), (5600, 1021), (3100, 1021), (3100, 2622), (3000, 2622)]

XY3

LP_E2.a → free_end((3350, 2820), -90, 6)

cpw 5862

[(5400, 3000), (5400, 1220), (3300, 1220), (3300, 2620), (3350, 2620)]

Z3

LP_E3.a → Q3.flux

cpw 6661

[(5250, 4500), (5250, 1422), (3550, 1422), (3550, 2622), (3450, 2622)]

XY4

LP_N3.a → free_end((3800, 2820), -90, 6)

cpw 7303

[(4500, 5320), (5100, 5320), (5100, 1620), (3750, 1620), (3750, 2620), (3800, 2620)]

Z4

LP_N2.a → Q4.flux

cpw 7802

[(3000, 5122), (4950, 5122), (4950, 1822), (4000, 1822), (4000, 2622), (3900, 2622)]

Hanger geometry: the feed travels +x, the resonators hang below (right), and the tee’s leg is at the downstream end with the short upstream, which is why direction="backward". s is measured from CAP_IN.b at x = 563.5. ground_gap 4 = |Δy| 26 − 5 − 6 − 5 − 6. The control lines in the source are laid out without crossings. Use r=20 (Metal fillet 20) on the short jogs.

Launchers, pads, test structures, markers, labels

There are 12 pads, 3 per edge on a 1.5 mm pitch. The feed enters at LP_W3 and leaves at LP_N1. XY: W2, S1, S3, E2, N3. Z: W1, S2, E1, E3, N2. There are no markers or labels.

LOM node structure

  • Q0–Q4: Xmon islands, each with a JJ to ground. Nearest neighbours share a direct island–island C. That coupling is geometric, so it needs a FEM C-matrix; the component estimates miss it.

  • R0–R4: λ/4 resonators, open at the claw (C_g) and shorted at the hanger, with a coupled-line κ to FL_P.

  • FL_P: a λ/2 Purcell filter (~10.0 mm) between CAP_IN and CAP_OUT (series C), which go to the 50 Ω ports LP_W3 / LP_N1.

  • Control: XYi is an open drive line with C_d to island i (port LP_). Zi* is a flux line to the junction (flux port). Use a SQUID (TrilayerSquid) if flux tuning is modelled.

Targets (if stated)

The only stated design choice is the staggered readout lengths, 4.41–4.57 mm in 0.04 mm steps, for distinct multiplexed tones. No Lj/Cj is set, so take the Metal default Lj 10 nH: junction={"Lj_nH": 10, "Cj_fF": 2}.

Simplifications for a first rebuild

  1. Build the feed as three routes (FL_IN, FL_P, FL_OUT) with the Purcell loop nodes. Hang all five resonators on FL_P.

  2. Use FingerCapacitorSquare for both CapNInterdigital caps. They are shorter (73 µm) than Metal’s 146 µm, so centre them at the Metal midpoints given above.

  3. Make each XY line a free end with term=6. Route each Z line to Xmon.flux (port end, so the tip shorts to ground).

  4. The ground pockets overlap at 450 µm pitch. If the DRC flags the merged gaps, keep the pitch (it carries the direct coupling) and waive the warning, or raise the pitch to 480 µm.

Quantcore PDK rebuild

The notebook is a Quantcore PDK rebuild of the source, not a replica: the netlist, LOM and imported components are kept, and the process is Quantcore’s.

Notebook: examples/designs/metal/m52_barends_5q.ipynb.

  • Chip: the Quantcore die frame (L dicing markers 1000/0, 3 × 3 alignment grids 1001/0, chip boundary 100/0, device ID where an edge is free; any brand text is QCT), on the 7.5 × 7.5 mm Quantcore die; the source’s 6 × 6 mm chip area is centred in it (box=(-750, -750, 6750, 6750)), so every coordinate stays the source’s (the source’s 6 × 6 mm, drawn 6.2 × 6.2 mm with the overhanging pads, fits).

  • Stack: sapphire 650 µm, εr 10.25 (the PDK default stack, used by the LOM estimates and every FEM export); Metal’s DesignPlanar default is silicon (εr 11.45, 750 µm), so the LOM capacitances, resonator frequencies and FEM jobs here are for sapphire rather than the source’s substrate.

  • Junctions: Quantcore Nb/AlOx/Nb trilayer junctions (junction_type from layouts.registry.trilayer_junction): fingers on 2/0 + 4/0, spacer 50/0 with the 51–53/0 gradient, sized on the PDK junction rule (Nb Δ = 1.4 meV, I_c·R_n = πΔ/2e, R_n·A 13.5 kΩ·µm² and C_J 50 fF/µm² assumed); the five TransmonCrossFL (XY + Z lines) each get a PDK SQUID of two 0.25 × 0.6 µm junctions → R_n = 45 kΩ in parallel, L_J = 6.73 nH at zero flux, C_J = 15 fF (in the 40–50 kΩ window) (the source has no hfss_inductance and leaves the junctions out of its GDS). Expected f_q ≈ 6.3 GHz (an estimate), close to the 6.2–6.4 GHz readouts, so the rebuild’s frequency plan needs the Z lines (or a resized junction) to detune them. The five qubits stay identical.

  • Kept: the topology and every Metal component with the source’s options (through the metal: wrapper), positions and orientations, every route’s end pins and total length, the sim groups (the feedline with its Purcell section, the five λ/4 readouts 4410–4570 µm, the ten XY / Z lines, the d.couple arm-to-arm estimate).