sign in sign up
<<
Home , Majorana fermion

PITT-PACC-1709

Baryogenesis from Oscillations of Charmed or Beautiful

Kyle Aitken,1, ∗ David McKeen,2, † Ann E. Nelson,1, ‡ and Thomas Neder3, § 1Department of Physics, University of Washington, Seattle, WA 98195, USA 2Pittsburgh Physics, Astrophysics, and Cosmology Center, Department of Physics and Astronomy, University of Pittsburgh, PA 15260, USA 3AHEP Group, Instituto de Física Corpuscular — C.S.I.C./Universitat de València, Parc Científic de Paterna, C/ Catedrático José Beltrán 2 E-46980 Paterna (Valencia), Spain (Dated: August 7, 2017) We propose a model for CP violating oscillations of neutral, heavy-flavored baryons into an- tibaryons at rates which are within a few orders of magnitude of their lifetimes. The flavor structure of the violation suppresses oscillations and baryon number violating nuclear decays to experimentally allowed rates. We also propose a scenario for producing such baryons in the early Universe via the out-of-equilibrium decays of a neutral particle, after hadronization but before nucleosynthesis. We find parameters where CP violating baryon oscillations at a temperature of a few MeV could result in the observed asymmetry between baryons and antibaryons. Furthermore, part of the relevant parameter space for baryogenesis is potentially testable at Belle II via decays of heavy flavor baryons into an exotic neutral . The model introduces four new : three light Majorana and a colored scalar. The lightest of these fermions is typically long lived on collider timescales and may be produced in decays of bottom and possibly charmed .

PACS numbers: 11.30.Fs,14.20.-c,14.80.Ly,14.80.Nb

I. INTRODUCTION a high reheat temperature or production of heavy parti- cles during reheating means a low baryogenesis scale is The puzzle of baryogenesis, how the Universe came consistent with a wider variety of inflationary models [6]. to be composed primarily of rather than equal Lower reheat temperatures are possible provided the amounts of matter and antimatter, has led to numerous inflaton decays produce heavy particles which decay out theories about physics beyond the (SM), of thermal equilibrium in a baryon and CP violating man- beginning with the pioneering work of Sakharov [1]. ner [7]. In Ref. [7] baryogenesis occurs due to the baryon Three ingredients are present in one form or another in number violating decays of TeV mass squarks in an R- any baryogenesis theory: baryon number violation, C and violating supersymmetric model, in which the re- CP violation, and departure from thermal equilibrium. heat temperature could be as low as an MeV, provided Because baryon number violation is required, initially that the heavy squarks can be produced out of equi- baryogenesis was thought to involve new baryon number librium at the end of inflation. Such squark mediated violating processes which are only important at very high baryon number violation is consistent with the observed energies, although it was later realized that anomalous lifetime of the , due to the conservation of lep- electroweak processes could do the job at temperatures ton number, and, depending on the flavor structure of as low as the weak phase transition [2]. the baryon number violating operators, can be consis- Most baryogenesis models require the Universe to re- tent with the stability of heavy nuclei as well. In Ref. [8] heat after inflation to a high temperature, typically well it was pointed out that such heavy flavor baryon number above the weak scale. However, many theories of physics 0 violation could lead to oscillations of the Ξb baryon at beyond the SM are inconsistent with a high inflation scale a rate comparable to its lifetime, while being consistent or are inconsistent with a high postinflation reheat scale.

arXiv:1708.01259v1 [hep-ph] 3 Aug 2017 with the lifetime of heavy nuclei. dark matter, if the axion is present during infla- Here we present a baryogenesis model which is consis- tion, requires a low inflation scale in order to avoid ex- tent with a reheat temperature as low as a few MeV, and cessive isocurvature perturbations [3]. which requires no postinflationary production of any par- requires a low reheat scale in order to avoid overproduc- ticle heavier than about 6—10 GeV. The required baryon tion of the [4]. The relaxion solution to the number violation is conceivably observable via the oscil- requires a low inflation scale so that lations of heavy-flavor neutral baryons, and the required the Hubble temperature during inflation does not sup- CP violation is potentially of in such oscillations. press instantons [5]. In addition, avoiding the need for (1) The processes that produce theO baryon asymmetry in the early Universe involve particles and phenomena which can be directly studied in the laboratory – a unique fea- ∗ [email protected] ture of our theory. Our proposal is that certain neutral † [email protected] heavy flavor baryons undergo CP and baryon number ‡ [email protected] violating oscillations and decays, and are produced in § neder@ific.uv.es the early Universe via the out of equilibrium decays of 2 a weakly coupled neutral particle whose lifetime is of or- jorana fermions, χ1,2,3, and a single color triplet scalar, der 0.1 s, a time when the temperature is of order a few φ. The interactions involving the new particles and weak MeV. The basic scenario was outlined in Ref. [9], and the SU(2) singlet SM are given by model we study has the same field content and couplings ∗ c c int gudφ u¯Rd yidφχ¯id + h.c., (1) as Ref. [10]. The basic formalism for analyzing CP vio- L ⊃ − R − R lation in fermion antifermion oscillations was worked out along with terms involving other generations, d s, b in Ref. [11]. and u c, t. By convention we take all two component→ The outline of the paper is as follows. In section II the fields to→ transform in the left-handed representation un- model is introduced, and the effective operator responsi- c der Lorentz transformations. dR stands for the charge ble for baryon oscillations is constructed. In section III, conjugate of the right-handed down field, which is general ∆B = 2, six-quark effective operators are ana- in the left-handed Lorentz representation. lyzed for their contribution to dinucleon decay, that is, The required new particles and their interactions are the decay of two into . Currently, dinu- motivated as follows. A natural way to construct the cleon decay places similar or stronger constraints on all ∆B = 2 six-quark operator we require for baryon oscilla- such operators than does neutron oscillations. For oper- tions is from two ∆B = 1 four-fermion interactions con- ators that cannot contribute to dinucleon decay at tree nected by an exotic neutral Majorana fermion. Thus we level, electroweak corrections to the six-quark operators introduce an exotic, electrically-neutral, colorless, Majo- are examined. In section IV, the general formalism for rana fermion, χ1, which couples to other fermions via a CP-violating oscillations of fermions is reviewed, and the 0 four-fermion interaction of the form uRdRdRχ1 (using u oscillation parameters are calculated for the model in- and d here to represent any up- or down-type quark). troduced in section II. In sections V and VI, direct con- Since such a four-fermion interaction is itself nonrenor- straints on the masses and couplings of the new φ and χ malizable, we also introduce a complex, color triplet, particles from collider searches, and indirect constraints scalar particle () φ to mediate the ∆B = 1 in- from rare decays of mesons and baryons are derived, re- teractions. Note that if χ1 is heavier than the difference spectively. Section VII contains our analysis of how in in mass between the proton and , mp me = this model the baryon asymmetry of the Universe (BAU) 937.76 MeV, this interaction does not give rise to− proton is produced. Finally, in section VIII we conclude and 1 decay. In the presence of only χ1, there is no physical point at possible directions for future work. CP violation, as there is enough reparameterization free- dom to remove the phases in the couplings. We introduce

a second fermion, χ2 (with mχ2 > mχ1 ), in order to give II. MODEL rise to CP violation. Finally, for baryogenesis, the oscil- lating baryons must be produced out of thermal equilib- We wish to find a theory which allows for sufficiently rium. As described in Sec. VII, this is most simply ac- large baryon number and CP violation to explain baryo- complished by introducing a third Majorana fermion, χ3, genesis at relatively low energy. In order to ensure suf- which decays out of equilibrium to produce the baryons ficient stability of the proton, we assume number whose oscillations result in baryogenesis. is not violated, other than perhaps via the tiny ∆L = 2 Note that we only consider operators constructed out terms that could account for Majorana masses. of right-handed quarks, for two reasons. Our phenomeno- The lowest dimension terms which violate baryon num- logical reason is that, as we will show in Sec. III, right ber and not are dimension 9, six-quark handed quark operators are less constrained by dinucleon ∆B = 2 operators. Such operators can lead to neutral decay due to the requirement of light quark mass inser- baryon oscillations and conceivably CP violation [9], and tions in flavor-changing loops. Our top down theoretical can arise as an effective field theory description of physics reason is that, as mentioned, the interactions in Eq. (1) at some higher energy scale. A minimal renormalizable occur in RPV SUSY models, suggesting a possible em- model for generating such terms involves a new charge bedding of our model into a more complete theory. 1/3 color triplet scalar and two Majorana fermions, as Figure 1 shows the ∆B = 2 six-quark operators that described− in Ref. [10]. A third Majorana fermion, which are generated by the interactions in Eq. (1). Such oper- decays out of thermal equilibrium, allows for the fulfill- ators, which can mediate the transition of a baryon to 0 B ment of the out-of-equilibrium Sakharov condition. an antibaryon ¯ , can be written as B We note that this model for baryon number violation abc def 0 =   can easily be embedded in an R-parity violating super- BB O h i symmetric (RPV SUSY) theory. In such theories, the i j 0 0 k 0 (qR)a (qR)i,b (qR)c (qR)j,d (qR)e (qR)k,f + ... , would play the role of the Majorana fermions × (2) and a down-type SU(2) singlet squark can be the colored scalar. For simplicity, we do not explore this embedding in a SUSY framework in this paper and we stick to the minimal version of the model. 1 The stability of 9Be leads to a marginally stronger lower bound Our model thus adds four new particles: three Ma- of mχ1 > 937.9 MeV [9]. 3

a common decay product, through 0 0 and ¯0 0 u u¯ Ξ Λ Ξ Λ R R (ignoring any neutral products).→ However, we→ will find such processes are suppressed relative to their direct φ mχ φ oscillation cousins, so we will ignore them in our analysis. ¯ dR dR There are also baryon-number–preserving operators that χ contribute to the masses and mixings of SM baryons. These are greatly suppressed relative to those that occur in the SM and we do not consider them further either. dR d¯R Therefore, in what follows, we focus on operators BB with coefficients of the form of Eq. (4). O FIG. 1. The basic six-quark ∆B = 2 operator generated by φ and χ exchange. u and d here represent any of the up- or down-type quark flavors. The quarks involved are all weak SU(2) singlets, as emphasized by the R subscripts. III. DINUCLEON DECAY CONSTRAINTS

As described in the preceding section, we would like where a, . . . , f are color indices, i, j, k spinor indices, and our six-quark operators to allow for the oscillation of q = u, d, s, c, b any of the quark flavors (because of its heavy baryons in order to produce the Universe’s ob- short lifetime, the does not hadronize and is served baryon asymmetry. In Sec. VII, we will show that not important in the low-energy effective theory) which the ideal width for such an oscillation, which is depen- are all right-chiral. The ellipsis represents other possible dent upon the value of CBB in Eq. (4), is a few orders of permutations of color or spinor indices. Here, denotes magnitude smaller than ’s decay width. However, mod- B an arbitrary standard model baryon with the quark con- els with B violation by twoB units are certainly not a new tent qqq while 0 contains q0q0q0. [We will use both the idea, and so significant experimental effort has been put B baryon name or the quark content (qqq) to label the fourth into constraining processes. The most B ∆B = 2 operators in question throughout this paper.] The pre- immediate constraint on our six-quark operators is the cise index structure of BB0 is not important for the pur- lack of observed dinucleon decay, which we quantify in O poses of this paper. Therefore, in what follows, we will this section. The analysis we perform here applies to suppress the indices on BB0 and generically denote the six-quark operators in general and is independent of the O operators we are interested in that appear in the effective origin of the new physics introduced in Sec. II. Lagrangian via the shorthand Dinucleon constraints come from underground detec- 0 0 0 tors whose primary purpose is the detection of proton eff CBB0 (qqq)(q q q ) CBB0 BB0 , (3) L ⊃ ≡ O decay and neutrino oscillations. For example, in a nu- cleus, a n n¯ transition will be shortly followed by keeping in mind that the leading operators that are gen- → erated involve only right-chiral quarks. the annihilation of the n¯ with one of the other nucleons, Matching the interactions generated by Eq. (1) to the leading to the decay of the nucleus of mass number A to a nucleus with A0 = A 2 plus mesons. The lack effective theory at tree-level gives the coefficient of the − operator that generates oscillations between a neutral of observation of such decays can therefore bound the baryon and its , ¯, transition amplitude δnn [13] which is related to the co- 2 B ↔ B efficient of the (udd) operator, Cnn [cf. Eqs. (4) and 2 16 ∗ ∗ ! (5)]. Currently, the lower bound on the O lifetime X mχ g yid0 + g 0 yid i ud ud (in decays to ) of 32 from the Super- CBB 2 2 2 , (4) 1.9 10 years ∼ m mχ m × i B − i φ Kamiokande collaboration [14] places the strongest limit, −33 δnn < 1.9 10 GeV. with , , and 0 labeling the quarks comprising . For × u d d Operators that also violate strangeness do not directly example, the operator (ddc)2 would allow the processesB induce n n¯ transitions in a nucleus. However, they Σ¯ Σ . Given this operator, we will find it useful to c c can also lead→ to dinucleon decays, A (A 2)+mesons, relate↔ the coefficient to the (dispersive) transition ampli- through the reaction NN → + X −where N is a tude, defined by δBB ¯ CBB BB , with → ≡ hB| O |Bi . For example, the diagram on the left of Fig. 2 2 shows how the operator (uds)2 can lead to dinucleon de- δBB = κ CBB, (5) cay to a pair of kaons. The Super-Kamiokande collabo- where κ 10−2 GeV3 [12]. In analogy with meson oscil- ration [15] has searched for such decays and has placed lations, when∼ the two-state system in question is unam- an upper bound on the pp K+K+ decay rate by lim- 16 → 14 + + biguous, δBB can also be referred to as M12. iting the lifetime for O CK K to more than 32 → Operators which involve different baryons of the form 1.7 10 years. × BB0 would allow for a common decay product between To make use of this limit, we start with the effective O and ¯ and could also give rise to oscillations. For operator BB. The dinucleon decay rate through direct Bexample,B (uss)(uds) would allow for Ξ0 and Ξ¯0 to have nucleon annihilationO can then be roughly approximated 4

lead to dinucleon decay. An illustration of this is shown K+/K0 u/d u/d on the right of Fig. 2. K+/K0 p/n p/n s¯ To properly estimate the rate for dinucleon decay from d d s¯ u ∆B = 2 operators (involving heavy flavors), we must u s 0 match the UV theory involving quarks to a low energy W ± u s¯ K u effective theory involving baryons valid at momentum d d s¯ u s¯ transfers below 4πfπ 1 GeV where fπ = 93 MeV. p/n + 0 p/n ∼ K /K u/d This consists of writing down an operator in the UV u/d K+/K0 theory and treating the coefficient of this operator as a spurion that transforms in a particular way under the global chiral quark flavor symmetry so 2 SU(3)L SU(3)R FIG. 2. Left: Dinucleon decay via the ∆B = ∆S = 2 (uds) as to make the operator invariant. This operator× is then 0 ¯ 0 operator that mediates Λ ↔ Λ oscillations. Right: Din- matched onto an operator in the effective theory that ucleon decay mediated by the ∆B = 2, ∆S = 4 (uss)2 transforms in the same way under the chiral symmetry operator that becomes ∆S = 3 (uds)(uss) operator in the presence of flavor-changing weak interactions. Because the with the same spurion coefficient. In the UV, the light short-distance ∆B = 2 operators we consider involve weak quarks qL,R transform as triplets under SU(3)L,R. In the isosinglets, this operator requires light quark flips, low energy theory, the meson octet, Π, is described by a indicated by crosses. See text for discussion of the match- field Σ = exp (2iΠ/fπ) which transforms under the chi- † ing of the short distance theory onto the (chiral symmetry ral symmetry as Σ LΣR where L, R are SU(3)L,R violating) long distance theory. transformations, respectively.→ Incorporating the baryon octet (see, e.g., Ref. [17]) can be done by defining a † field ξ = exp (iΠ/fπ) which transforms as ξ LξU , † → by considering the decay rate to a meson pair [16], ξ UξR under SU(3)L,R. U is an SU(3) matrix that → 2 depends nonlinearly on the meson fields. The baryon 9 CBB 2 octet B is defined to transform as B UBU †. Oper- ΓNN→X | 2 | 2 mesons BB NN ρN (6) → ∼ 32π mN |h |O | i| ators in the effective theory are then constructed out of Σ, B, and ξ along with spurions from the UV theory to −3 where mN is the nucleon mass, ρN 0.25 fm is the be invariant under the flavor symmetry. Since the chiral ' nucleon density, and we have ignored the masses of the symmetry is dynamically broken by the strong coupling final state particles. In the case of operators that can con- of QCD around 4πfπ, one can use naive dimensional anal- tribute at tree level, the matrix element can be estimated ysis to properly account for factors of 4π (that come from 5 5 as roughly 2 mesons BB NN ΛQCD (200 MeV) . the strong coupling) and the cutoff, , that appear in h |O | i ∼ ' 4πfπ Using this and Eq. (5), the limit on the rate for NN this matching procedure, as described in, e.g., Ref. [18]. KK from Super Kamiokande translates to a limit on the→ We will first illustrate this matching procedure in our transition amplitude of theory with interactions given by Eq. (1), assuming for now that only the light quarks , , and are involved. −30 u d s δ(uds)2 . 10 GeV. (7) We will deal with heavy quarks c and b below. Since distinction between chiralities is necessary, we will tem- In what follows, we also take operators that change porarily denote them explicitly. After integrating out strangeness by one or three units to have roughly the the scalar, φ, and the Majorana fermions, χ , we are left same bound as this. i with a ∆B = 2 operator involving only (light) right-chiral Kinematic constraints protect certain operators from quarks, C (q q q )2. This operator must be matched contributing to dinucleon decay at leading order. Op- BB R R R onto an operator valid at long distances involving baryons erators such as 2 that change strangeness by four (uss) at the scale of chiral symmetry breaking. The coeffi- units (i.e. ∆S = 4) are kinematically forbidden from con- cient C can be treated as a spurion that transforms tributing to dinucleon decay at tree level since 2m < BB N under SU(3) in a representation that appears in the . Similarly, those that involve charm2 or bottom R 4mK tensor decomposition of 6 triplets. For definiteness, take quarks also do not lead to dinucleon decay at leading or- it to transform as an SU(3) octet. Then the object der. However, when combined with flavor-violating weak R ˜ † ˜ ˜ † interactions, these operators involving heavy quarks can CBB ξCBBξ transforms as CBB UCBBU and the operator≡ matching is →

2 32 CBB(qRqRqR) 4πf trBC˜BBB + ..., (8) → π 2 Depending on the nucleon binding energy, nn → Dγ through where the ellipsis represents other possible orderings of a ∆C = 1 operator is kinematically allowed for some nuclei, the baryon octets and the spurion. Note that this gives but due to the dependence of the amplitude on the mo- 3 −2 3 mentum and coupling and phase space suppression the rate is an understanding of the size of κ 4πfπ 10 GeV in 3 −9 ∼ ' proportional to (α/4π)(kγ /mN ) ∼ 10 , where kγ is the pho- Eq. (5). Adjusting this analysis if CBB transforms under ton energy, suppressing the rate below other decays with less a different representation of SU(3)R is straightforward; constrained phase space. one inserts the required numbers of ξ and ξ† into the 5 definition of C˜BB so that it transforms in such a way as dL u¯LuR u¯LuR dL to leave trBC˜BBB invariant. For example, if CBB is a ⟨ ⟩⟨⟩ singlet then one simply takes C˜BB CBB. Four-quark weak operators involving≡ light quarks can be matched onto the low energy effective the- ory in much the same way. The coefficient of the j operator u¯ γµq q¯ γ u can be viewed as a spurion uL uL L L Li µ L bb that transforms as an octet under SU(3)L and the strangeness changing (∆S = 1) coefficient takes a value ∗ i i 3 † † † GF VusVudh with h j = δ2δj . Then ξ hξ Uξ hξU ∝and the matching is →

dR dR GF ∗ µ 5 ¯ 5 VusVuduγ¯ 1 γ sdγµ 1 γ u √2 − − (9) FIG. 3. The ten-quark ∆B = 2 operator that results from GF 2 ∗ 3 ¯ † the leading (uRdRbR) operator after integrating out the b VusVud 4πfπ trBξ hξB + ..., → √2 quarks. The crosses represent chirality-flipping b quark mass insertions. We use hu¯LuRi to indicate the pairs of light quark where again the ellipsis represents other possible order- fields that can be replaced by the chiral condensate when ings of B, B¯, and ξ†hξ. matching onto the long distance theory relevant for dinucleon Now we can combine a ∆B = 2 operator that also decay. changes strangeness by n units with the weak ∆S = 1 operator to form a ∆B = 2, ∆S = n 1 operator that is − given by 2 means that the induced ∆B = 2 operator (uLdRdL) is suppressed relative to (u d b )2 by the factor GF ∗ 2 32 ¯ ˜ † R R R VusVudfπ 4πfπ trBCBBξ hξB + .... (10) √2  3 2 In other words, if the leading ∆B = 2 operator has ∆S = GF ∗ 4πfπ −20 VubVud 10 . (14) n, the ∆B = 2, ∆S = n 1 operator that is generated √2 mb ∼ due to weak interactions− is suppressed relative to it by the factor Additionally in the case of a leading operator containing 2 GF ∗ 2 −8 a b and c quark, e.g. (dRcRbR) , there are perturbative VusVudfπ 10 . (11) √2 ∼ loops that generate dimension-nine operators involving light quarks above the chiral symmetry breaking scale. 2 Thus, for example, the bound on a leading ∆S = 4 op- In the case of the operator (dRcRbR) , two such loops 2 2 erator (uRsRsR) from the lack of dinucleon decay is can be used to generate the operator (uLdLdR) with a around eight orders of magnitude weaker than that on coefficient suppressed relative to the leading one by the ∆S = 3 operator (uds)(uss) [which we take to be comparable to that on the ∆S = 2 operator (uds)2],  2 2 GF ∗ mbmc mW −16 −22 VubVcd log 10 . (15) δ(uss)2 10 GeV. (12) 2 2 . √2 4π mb ∼ Now, we consider the case where the leading ∆B = 2 operators contain heavy quarks. Consider, for example, In Table I, we list operators that can mediate if after integrating out the heavy scalar φ and Majo- ¯ transitions along with the number of loops requiredB ↔ rana fermions χi, that the leading operator we generate B 2 for each operator to mediate (∆S = 0, 1, 2, 3) dinucleon is C(udb)2 (uRdRbR) . Before matching onto the theory decay. We show the resulting limits on the transition valid after chiral symmetry breaking we must first inte- 2 amplitudes δBB = M12 = κ CBB of each operator from grate out the b quarks. In the presence of weak interac- the lack of observation| | of dinucleon decay, accounting tions, as shown in Fig. 3, doing so will lead to a ten-quark for the appropriate suppression factors. In general we operator, find that only operators which require 2 or more weak  2 interactions to contribute to dinucleon decay can give 2 GF ∗ 1 baryon oscillations at a rate which is large enough to be C(udb)2 (uRdRbR) VubVud → √2 mb (13) relevant for either experimental searches or baryogenesis. 2 C(udb)2 (uRdRdLu¯LuL) . The last column of the table gives the limit on the size of × the operator that can be produced in our specific model After chiral symmetry breaking, u¯LuR can be replaced when collider constraints on new particles are considered, 3 by the quark condensate which is roughly 4πfπ. This which will be discussed in Sec. V. 6

Weak Insertions Measured Limits on δ = M (GeV) Operator B BB 12 Required Γ (GeV) [19] Dinucleon decay Collider (udd)2 n None (7.477 ± 0.009) × 10−28 10−33 10−17 (uds)2 Λ None (2.501 ± 0.019) × 10−15 10−30 10−17 (uds)2 Σ0 None (8.9 ± 0.8) × 10−6 10−30 10−17 (uss)2 Ξ0 One (2.27 ± 0.07) × 10−15 10−22 10−17 2 0 +0.11 −3 −17 −16 (ddc) Σc Two 1.83−0.19 × 10 10 10 2 0 +0.58 −12 −16 −15 (dsc) Ξc Two 5.87−0.61 × 10 10 10 2 0 −12 −14 −15 (ssc) Ωc Two (9.5 ± 1.2) × 10 10 10 2 0 −13 −13 −17 (udb) Λb Two (4.490 ± 0.031) × 10 10 10 2 0∗ −3∗ −13 −17 (udb) Σb Two ∼ 10 10 10 2 0 −13 −10 −17 (usb) Ξb Two (4.496 ± 0.095) × 10 10 10 2 0 † −12† −17 −15 (dcb) Ξcb Two ∼ 10 10 10 2 0 † −12† −14 −15 (scb) Ωcb Two ∼ 10 10 10 2 0 ‡ −13‡ −17 (ubb) Ξbb Four ∼ 10 >1 10 2 0 † −12† −15 (cbb) Ωcbb Four ∼ 10 >1 10

TABLE I. Operators that mediate B ↔ B¯ oscillations and the number of weak interaction insertions required for each of these to contribute to dinucleon decay. The resulting limit from dinucleon decay on the transition amplitude, defined in Eq. (5), for each operator is shown. An ∗ indicates a baryon that has not yet been observed and which has a strong decay channel open. A † (‡) indicates an unobserved baryon which primarily decays through a weak interaction of a c (b) quark.

IV. CP VIOLATION IN HEAVY BARYON with OSCILLATIONS  1 1  1 −imHt− ΓHt −imLt− ΓLt g±(t) = e 2 e 2 . (20) The evolution of the ( , ¯) system in vacuum, assum- 2 ± B B ing CPT conservation, can be described [9] by a 2 2 A particularly useful quantity that measures the level of × Hamiltonian, CP and baryon number violation is the quantity,

i i ! i MB ΓB M12 Γ12 PB→B PB→B¯ + PB→B¯ PB→¯ B¯ 2 2 AB = − − , (21) = M Γ = ∗ − i ∗ − i . (16) H − 2 M Γ MB ΓB PB→B + PB→B¯ + PB→B¯ + PB→¯ B¯ 12 − 2 12 − 2 M and Γ are both Hermitian matrices that describe the where, e.g., PB→B¯ is the time integrated probability for ¯ ¯ an intitial state to oscillate into a ¯ and the other dispersive and absorptive parts of the , , am- B B plitude, respectively. This system is entirelyB B → analogous B B terms are defined analogously. In terms of the elements of , this can be concisely expressed, to the very well known case of neutral mesons and an- H timesons. Because of the off-diagonal terms in , the ∗ H 2Im (M12Γ12) mass eigenstates L,H with masses mL,H are linear com- A = . (22) |B i B 2 2 binations of the flavor eigenstates and ¯ , ΓB + 4 M12 |Bi |Bi | | This expresses the familiar fact that CP violation requires L,H = p q ¯ . (17) |B i |Bi ± |Bi a phase difference between the absorptive and dispersive The mass difference is ∆m = mH mL > 0 and the width parts of the transition amplitudes. − difference between the states is ∆Γ = ΓH ΓL and can be The dispersive part of the transition amplitude, M12, − of either sign. The flavor admixtures can be determined is dominantly given by off-shell χi exchange in our model, by as seen in Fig. 1. We have already written down what we need to estimate this in Eqs. (4) and (5), resulting in  2 ∗ ∗ q M12 (i/2)Γ12 = . (18) 2 − ∗ ! p M12 (i/2)Γ12 mχ g y 0 − 2 X i ud id M12 κ 2 2 2 . (23) ∼ m mχ m A state that begins at t = 0 as a or ¯ is at time t i B − i φ |Bi |Bi 0 q ¯ Here, u, d, and d refer to the flavors that comprise (t) = g+(t) g−(t) , 0 ∗ ∗ B |B i |Bi − p |Bi and we have assumed, if d = d , that gudyid0 gud0 yid. p (19) If we concentrate on the contribution6 due to a particular ¯(t) = g+(t) ¯ g−(t) |B i |Bi − q |Bi χi and express it in terms of its mass difference from the 7 baryon, ∆mBi = mB mχ , we have χ1 to its contribution to M12 is − i 2  2 2 ∗ ∗ Γ12 ∆mB1 0 κ gudyid + gud0 yid 4π M12 i 2 M12 1 ∼ mB | | ∼ 2∆mBi m (29) φ  2  2   ∆mB1 5 GeV −16 500 MeV 0.1 . 8 10 GeV (24) ' 500 MeV mB ' × ∆mBi !4 The CP-violating quantity in Eq. (22) linearly de- 600 GeV AB pends on Γ . Without finely tuning the contributions p ∗ ∗ . 12 × mφ/ g yid0 + g 0 yid | | | ud ud | due to χ1 and χ2 against each other, this value of the ratio due to χ1 alone is roughly as large as the total ratio The absorptive part of the transition amplitude re- Γ12/M12 can get. quires an on-shell state into which both and ¯ can | | B B decay. This requires at least χ1 to be light enough for either baryon or antibaryon to decay into it. CP viola- V. COLLIDER CONSTRAINTS tion will be largest when the mass splitting between χ1 and is not too large. In this case the most important To obtain a large amount of CP violation in heavy B states for Γ12 are decays of to χ1 plus a meson. The baryon oscillation, it will be clear that the lightest two 0 B contribution from χ1π , for instance, can be estimated Majorana fermions must have masses on the order of a using the effective Lagrangian, few GeV along with couplings to quarks that are not too small. In this discussion, we consider the two lightest Ma- 0 ¯ 5 eff yiBπ iγ χi + h.c., (25) jorana fermions. The third, χ must be weakly coupled L ⊃ − B 3 in the minimal version of the model, due to cosmological where considerations as we will see in Sec. VII.

∗ The constraints that we will discuss in this section re- 4πκ gudyid0 quire that φ have a mass of at least a few hundred GeV. yiB 2 . (26) ∼ mB mφ In this case, χi decays can be analyzed by integrating out the scalar in Eq. (1) resulting in four-fermion inter- The factor of 4π in this expression accounts for the non- actions, perturbative of the interaction, which is similar to the -nucleon vertex. This interaction gives a con- gudyid0 c 0c 2 χ¯iu¯RdRdR + h.c. (30) tribution to Γ12 of − mφ

0 2 (For d = d , we have again assumed that gudyid0 X yiBmχi 1/2 0 0 6  Γ12 (1 + rχi rπ ) λ (1, rχi , rπ ) , 0 .) Both the interactions responsible for the de- ∼ 32π − gud yid i cay of the Majorana fermions and those that source (27) 2 2 2 2 2 CP-violating baryon oscillations are of the same form. where r 0 = m 0 /m and λ (a, b, c) = a + b + c χi,π χi,π B The quarks involved in the decay operator must be 2ab 2ac 2bc. The magnitude of the contribution to− − − lighter than those responsible for baryon oscillations and Γ12 from a particular χi is roughly the couplings responsible for decay must be relatively smaller, to avoid stronger dinucleon decay limits from 2 2 2 ∗ yiB 2πκ gudyid0 ∆B = 2 quarks involving light quarks. Γ12 ∆mBi i 2 2 The interaction in Eq. (30) allows for the decay χi | | ∼ 8π ∼ mB mφ udd0, where u, d, and d0 are up- and down-type quarks→  ∆m  1 10−16 GeV Bi (28) light enough for this to be kinematically allowed. It is ' × 500 MeV reasonable to assume that one mode dominates their al- !4 lowed branchings and in this case, their lifetimes are 5 GeV 2 600 GeV . p ∗ 3 2 2 × mB mφ/ gudyid0 m | | 2 (8π) φ τχi 5 ∼ m gud yid0 We now see the reason we need at least two Majorana χ1,2 (31)  5  4 fermions. If there were only a single Majorana fermion, 5 GeV mφ/√gud yid0 10−6 s . χ1, that contributed to M12 and Γ12, they would have ' mχi 20 TeV the same phase and AB in Eq. (22) would vanish. Thus, we use contributions from χ and χ exchange to obtain 0 1 2 For mχi = 5 GeV, with couplings gud yid . 2 a physical, CP-violating phase difference between M12 (mφ/20 TeV) the ∆B = 2 transition amplitude in 0 −22 and Γ12. the udd system is less than 10 GeV, avoiding con- The ratio of the single meson contribution to Γ12 from flict with constraints from dinucleon decay (see Table I). 8

0 Furthermore, if mχi = 5 GeV, as long as gud yid & or to χi plus a quark, 2 (mφ/350 TeV) , τχi . 0.1 s which is a short enough lifetime to avoid spoiling successful BBN (see, e.g., [20]). 2 X yidj Γφ→χd mφ, (33) ' 16π We might ask whether instead of ensuring that χ1,2 i,j decay fast enough to avoid spoiling BBN, the lightest fermion could instead be long enough lived to serve χ1 where we have assumed that mφ is much larger than as dark matter. First we note that the range allowed for the mass of any decay product. Therefore, these scalars kinematic stability of both proton and χ1 is extremely can appear in searches for dijet resonances (either singly fine tuned, with the χ1 mass between mp me and or pair produced) and (mono)jets and missing energy. . If we assume all the ’s participate− in a vi- mp + me χ Which search is most sensitive depends on mφ and the able heavy flavor baryogenesis mechanism, we will see branching fractions for φ u¯d¯ and φ χd. that we must require the χ1,2 masses to be around 3– Taken together, LHC searches→ for pair→ produced dijet 5 GeV, and we also need sufficiently large four-fermion resonances, both with [22] and without [23] heavy flavor interactions involving χ1,2 and heavy flavor quarks, sup- in the final states, as well as standard SUSY searches pressed by a scale Λheavy mφ/√gy 600 GeV. Here g ≡ ∼ for (b-tagged [24, 25] or not [25, 26]) jets plus missing en- and y here label couplings with the relevant flavor struc- ergy rule out φ masses below about 400 GeV. Above this ture. Based on our discussion in Sec. III, at the one loop mass, limits from pair produced dijet resonances are no level we must generate four-fermion operators involving longer constraining while resonant production of a sin- light quarks (into which χ1,2 can decay) are generated 2 4 6 gle φ with a rate proportional to gud for some u, d with a scale Λlight 10 Λheavy 10 GeV. This pro- | | & ∼ is important [27]. We use the limits from resonant di- vides a lower bound on the strength of the light quark jet production from [27] and recast searches for jets and four-fermion operator which gives an upper bound on missing energy [24–26] as well as monojets [28] to find the χ1,2 lifetime of τχ1,2 1000 s in the absence of fine- . limits on the couplings gud and yid0 as functions of mφ. tuning against some other source of this operator. There We find this limit for every flavor u, d, and d0, assuming is also an unavoidable decay channel that comes from the that only gud and yid0 are relevant. Given these limits, mixing of χ1,2 with the heavy flavor baryon, , whose the maximum value of the product of couplings gudyid0 oscillations are responsible for the BAU, withB a mixing at each mφ can be found, and taking a value of the mass angle θ M12 /∆m where ∆m is the mass splitting be- 0 splitting between χi and the udd baryon, which can be tween the∼ | Majorana| fermions and . This mixing leads B turned into an upper limit on the transition amplitude to the decay of χ into ’s decay channels with a partial 0 3 1,2 0 in the system. We show the upper 2B δudd = M12 udd width proportional to θ ΓB, which is much shorter than limit on M12 as a function of mφ for each pattern of the lifetime of the universe. flavors u, d, and d0, assuming the dominance of one par- We see that the Majorana fermions are generically un- ticular pair of couplings gud, yid0 and a mass splitting 0 stable but long-lived on the scale of collider experiments between mχi and the udd baryon of 200 MeV in Fig. 4. and appear as missing energy. Decay lengths on the or- We also show the largest value of M12 allowed from col- der of 102 to 107 m are expected, potentially relevant for lider searches in each neutral baryon system in Table I. the recently proposed MATHUSLA detector [21] which is optimized to search for long-lived particles. In what follows, to analyze collider constraints on the new scalar φ we will assume that any χi produced at a collider is VI. PHENOMENOLOGY invisible and defer discussion of the displaced decay sig- natures at, e.g., MATHUSLA. A. Hadron decays

Now that we know that the χi’s are invisible at col- liders, we can understand how the scalars appear when After integrating out the heavy colored scalars, four- produced in hadron collisions. Because φ is a color fun- fermion interactions between the Majorana fermions and damental, if it is kinematically accessible, φφ∗ pairs are quarks are generated as in Eq. (30). These can lead to easily produced in proton-(anti)proton collisions, and the new decays of hadrons to final states that differ in baryon signatures are essentially those of squarks in RPV SUSY. number by one unit along with any kinematically acces- In addition to QCD production, (single) scalars can be resonantly produced in the presence of some nonzero gud. Once produced, the scalar decays through one of the in- teractions in Eq. (1), either to quark pairs with a rate 3 Note that we perform this scan for a single Majorana fermion. Including a second, as we must to obtain CP violation, does not change the allowed values by more than an O(1) factor which, 2 g given our level of precision, is unimportant. The third, χ3 must X uidj be more weakly coupled than χ and can be even more safely Γφ→u¯d¯ mφ, (32) 1,2 ' 16π neglected here. i,j 9

where u and d here are the actual up and down quarks. (We have omitted the contribution to this operator from gudy1b which is more constrained by collider searches.) The rate for the quark to decay through such an inter- ��-�� b action is �������� 2 -�� 4 !  5  �� mb∆m guby1d ∆m Γ ¯ + �������� b→χ1u¯d ∼ 3 m2 O m5 | ( ��� ) �������� 60 (2π) φ b

�� (36) -�� 4 4 | � �� � �     �� ���� −15 ∆m 1.2 TeV �������� 2 10 GeV . ' × 2 GeV mφ/√guby1d � � ��-�� �� ���� In this expression ∆m is the mass splitting between χ1 and the (we have ignored masses in -�� �� the final state besides mχ1 ). We have chosen to nor- malize this expression on values of the mass splitting ������ and mφ/√guby1d that result in a transition amplitude -�� �� ������ −17 0 of M12 10 GeV in the Λb = (udb) baryon system, which| is| ∼ the rough collider limit. Given this mass split- | ( ��� ) ������

�� + -�� ting, this can lead to decays of B mesons to a nucleon | � �� ������ to χ1 with a branching ratio of

������ 4 -��   �� � � −3 ∆m �� �� Br ± 6 10 B →Nχ1+X ∼ × 2 GeV ��� ��� ��� ���� ���� ���� (37)  1.2 TeV 4 �ϕ (���) , × mφ/√guby1d

FIG. 4. Upper limits on M12 as functions of mφ that re- where X represents possible additional pions. This is not sult from collider searches for diject resonances and jets plus a small branching fraction, although final states of this missing energy, assuming the dominance of the product of form have not yet been searched for in decays. (0) couplings gudyid0 indicated, where u and d label generic However, the requirement that the final state hadrons up- and down-type quarks, respectively. We have taken carry baryon number means that this decay is kinemati- ∆mB = mB − mχ = 200 MeV. Top: The limits when yid or ± cally forbidden if mχ1 > mB mp = 4.34 GeV. Decays yis are dominant. Bottom: The limits when yib is dominant. − Solid curves show the limits in the case where the charge 2/3 of bottom baryons would be allowed to proceed for split- quark involved is u while dashed lines show the limit in the tings down to mπ, and one could expect branching ratios −3 4 case of the c quark. on the order of 10 for the parameters in Eq. (37). We also expect “wrong sign” decays of heavy baryons in this model, following a ¯ oscillation, with a branch- B → B sible χi, e.g., ing fraction that is roughly

meson baryon + χi [+ meson(s)], 1 M 2 → (34) 12 (38) baryon meson(s) + χi. | 2 | . → 2 ΓB As we showed in Sec. V, on the scale of 0 experiments, χ appear as missing energy. Consider, e.g., the Ωc . Given the constraints that appear i in Table I, this branching could potentially be as large as For definiteness, let us focus now on four-fermion in- −7 −1 teractions that involve the b quark and the lightest Ma- 10 . The Belle II experiment hopes to collect 50 ab of e+e− data at √s = 10.56 GeV collecting about∼ 50 109 jorana fermion. This is potentially relevant to the case 0 × where baryons containing b quarks undergo CP-violating B meson pairs. If Ωc baryons are produced in 2% of B oscillations in the early Universe, producing the BAU; meson decays (comparable to the measured production operators involving heavy quarks are less constrained by of Λc baryons), then there would be a sample of about 9 0 ¯ 0 dinucleon decay and are therefore more promising candi- 10 Ωc ’s and Ωc ’s. Thus, there could be a few hundred dates, cf. Table I. Similar considerations apply for oper- “wrong sign” decays in the data sample. While this would ators involving lighter quarks. Consider, as a definite example, b decays through the operator 4 The calculation of the baryon decay rate to χ1 and a single meson guby1d is essentially the same as that of Γ12 in Sec. IV, modulo a factor χ1uRdRbR, (35) 2 of mb/mχ ∼ O (1). − mφ 1 10 be a challenging measurement, it is interesting that it is H is the Hubble parameter which is related to the total in principle observable at the next generation B-factory energy density, given current experimental limits. s s We mention here that baryon-number–violating decays 8π ρ 8π ρ + ρ H = rad χ3 , (41) of baryons along these lines have been searched for by 3 M 2 ' 3 M 2 the CLAS Collaboration [29]. The branching fraction for Pl Pl 0 −5 Λ KS + inv. is limited to less than 2 10 while 19 → + ×−7 where MPl = 1.22 10 GeV is the Planck mass. In that for Λ pπ¯ must be less than 9 10 which are × → 2 × the absence of χ decays, ρ and ρ simply redshift sensitive to the operator (uds) . While interesting, these 3 rad χ3 −18 like radiation and matter energy densities, respectively. limit δ(uds)2 = M12 to less than about 10 GeV, which is less strong than| | the limit on this operator from null The right-hand sides of these equations describe how χ3 searches for dinucleon decay. In light of the less stringent decays cause the energy density in matter to decrease limits from dinucleon decay on operators involving heavy while dumping energy into the plasma. flavor, it would be highly desirable for searches for ∆B = In addition to depositing energy in the plasma, some 2 decays of baryons with heavy quarks to be performed. of the χ3 decays produce baryons and antibaryons, and ¯, that can oscillate and decay, violating CP andB B. ForB this to occur, the temperature of the Universe needs to be below the QCD confinement temperature, B. Meson oscillations TQCD 200 MeV. On the timescale of the expansion of the' Universe, H−1, the (anti-)baryons produced this In addition to the decays described above, the new way rapidly oscillate and decay, producing a net B asym- interactions could lead to flavor-changing oscillations of metry. However, because of the presence of the plasma, neutral mesons. The limits from these processes on this with which they can interact, as well as their large anni- model were considered in Ref. [10]; we refer the reader hilation cross section, ¯ can decohere in this envi- to [10] and references therein for further details. ronment, suppressing theB ↔ asymmetryB that is generated. Avoiding these constraints requires a suppression of Properly accounting for this requires a density matrix particular combinations of flavor-violating couplings. For treatment, which has been used in a cosmological con- example, considering oscillations, given mφ & text for neutrino oscillations and oscillating asymmetric 400 GeV, ys1 and ys2 could be (1) provided yd1, yd2 . dark matter [30, 31]. Following Ref. [31] (see [32] for a 10−2. Similar considerations applyO for flavor-violating similar analysis in the context of baryogenesis-related os- combinations of the couplings gus and gud. The con- cillations), we can write the Boltzmann equations that straints on charm and bottom couplings from D and B govern the evolution of the number density of the - ¯ oscillations are less severe. system, B B From the model building point of view, flavor-changing meson oscillations can be naturally avoided, e.g. charging dn † Γ± + 3Hn = i n n [O±, [O±, n]] the scalar under a symmetry so that F1 F2 is conserved, dt − H − H − 2 −   where F1,2 label flavor quantum numbers. 1 2 σv ± n, O±nO¯ ± n (42) − h i 2 { } − eq

1 Γχ3 ρχ3 VII. COSMOLOGICAL PRODUCTION OF THE + Brχ3→BO+, 2 mχ BARYON ASYMMETRY 3 where the last term describes and ¯ production We now answer in detail the question of how the baryon B B through χ3 decay. Brχ3→B is the branching ratio for χ3 asymmetry of the Universe is produced in this model. to decay to or ¯. In this equation n and n¯ are density In addition to the CP and baryon number violation de- matrices, B B scribed above, a nonzero asymmetry requires a departure from thermal equilibrium. The simplest possibility for ! ! nBB nBB¯ nB¯B¯ nBB¯ this is to assume that χ3 is very weakly coupled. It is n = , n¯ = , (43) therefore long-lived and decays out of equilibrium, pro- nBB¯ nB¯B¯ nBB¯ nBB ducing the baryons that undergo CP- and B-violating oscillations. and neq is the equilibrium density of baryons plus an- At temperatures below , the equations that deter- tibaryons. is the Hamiltonian seen in Eq. (16). σv ± mχ3 H h i mine the radiation and χ energy densities are and Γ± are thermally-averaged annihilation cross sec- 3 tions and scattering rates on the plasma, respectively. is a matrix dρrad O± + 4Hρrad = Γχ3 ρχ3 , (39) dt ! dρχ3 1 0 + 3Hρχ = Γχ ρχ . (40) O± = . (44) dt 3 − 3 3 0 1 ± 11

The subscript of σv ± and Γ±, i.e. whether they appear which diverges at small scattering angle, θ 0. This h i → with O+ or O− in Eq. (42), is determined by the be- divergence is cut off at finite temperature by the inverse havior of the effective Lagrangian that gives rise to these photon screening length, mγ . Using this, the total cross interactions under charge conjugation of only the heavy section can be estimated as ¯ baryons, , eff eff . Interactions that do not  2  change signB ↔ areB saidL to↔ be ±L flavor-blind while those that 2 2 4E σsc 4πα µ log 2 (50) do are flavor-sensitive. For example, and ¯ can scat- ∼ mγ ter on light charged particles in the plasmaB throughB their where E is the electron energy. Taking E T , mγ magnetic moment, µ, which corresponds to a term in the ∼ ∼ eT/3 [33] and µ 1/(2mB), this gives effective Lagrangian of ∼ 2   iµ ¯ ν ρ πα 9 [γ , γ ] Fνρ. (45) σsc 2 log . (51) 4 B B ∼ mB πα Under ¯ this term changes sign, so the rate for B ↔ B The (flavor-sensitive) scattering rate is therefore scattering via the magnetic moment appears with O− in the Botzmann equation. Γ− = Γsc σsc (ne− + ne+ ) It is useful to work in terms of the quantities ∼ πα2  9  3ζ(3) log T 3 Σ nBB + nB¯B¯, ∆ nBB nB¯B¯, 2 2 ≡ ≡ − (46) ∼ mB πα × π (52) Ξ nBB¯ nBB¯ , Π nBB¯ + nBB¯ .  2  3 ≡ − ≡ 5 GeV T 10−11 GeV . In this basis the Boltzmann equations are ∼ mB 10 MeV  d  Γ ρ χ3 χ3 At temperatures above a few MeV, as is needed for BBN, + 3H Σ = Brχ3→B ΓBΣ dt mχ3 − this rate is larger than a typical heavy baryon width and (Re Γ12) Π + i (Im Γ12)Ξ therefore strongly affects the - ¯ oscillations. − When solving the BoltzmannB B equations, we take an 1h  2 2 2  σv + + σv − Σ ∆ 4neq annihilation cross section that is similar to that for − 2 h i h i − − pp¯ annihilation at low energies,  2 2 i + σv + σv − Π Ξ , h i − h i − σv + + σv − = 400 mb. (53)  d  h i h i + 3H ∆ = ΓB∆ + 2i (Re M12) Ξ + 2 (Im M12)Π, dt − We will find that only the total annihilation cross section   and not whether it is flavor-blind or -sensitive is impor- d  + 3H Ξ = ΓB + 2Γ− + σv +Σ Ξ tant, since Σ ∆, Ξ, Π. Furthermore the annihilation dt − h i rate is always much smaller than the scattering rate at + 2i (Re M12) ∆ i (Im Γ12)Σ, temperatures we are interested in, so its effect on the final   − d  asymmetry is subdominant and can generally be ignored. + 3H Π = ΓB + 2Γ− + σv +Σ Π dt − h i

2 (Im M12) ∆ (Re Γ12)Σ. A. Sudden Decay Approximation − − (47) Coherent oscillations from a flavor-symmetric state to Having removed the heavy baryons from the problem an asymmetric state proceed through Σ Ξ, Π ∆. Flavor-sensitive scattering and flavor-blind→ annihilation→ due to the short timescales in their system, the evolution equations are Eqs. (39), (40), and (59). These involve suppress Ξ and Π and therefore lead to decoherence. When they decay, and ¯ create states that carry only the radiation energy density, χ3 density, and the baryon number. TheB flavor-asymmetricB configuration baryon asymmetry. They can be simply studied using a contributes to the difference between the baryon and an- sudden decay approximation to gain a rough estimate of tibaryon number densities, the baryon asymmetry. We outline this estimate below. At some high temperature above m , we assume that   χ3 d χ3 was in thermal equilibrium with the plasma, fixing its + 3H (nB nB¯ ) = ΓB∆. (48) dt − number density for T . mχ3 to roughly The dominant interaction of and ¯ with the plasma B B 3 ζ(3) 3 is scattering on charged particles (mostly at nχ3 T . (54) ' 4 π2 T . 100 MeV) via the magnetic moment term in Eq. (45). The cross section for this at temperatures well As the Universe cools the energy density in χ3 and radi- below mB is ation are equal. This occurs at the temperature  2  dσsc 1 + sin θ/2 45ζ(3) = α2µ2 (49) 2 Teq = 4 mχ3 . (55) dΩ sin θ/2 2π g∗(T0) 12

+ g∗ is the effective number of relativistic degrees of free- Here, tdec is the time just after decay. The ratio of the dom and here it is evaluated at T0 & mχ3 . This corre- temperatures just before and after decay is determined + − sponds to the time by ρχ3 (tdec) = (1 + ξ)ρχ3 (tdec) so that s 45 M T (t− ) Pl dec −1/4 −1/4 (62) teq = 3 2 + = (1 + ξ) ξ , 16π g∗(Teq) T T (t ) ' eq (56) dec 7 2 1 π g∗(T0) MPl and = p 2 2 . 5πg∗(Teq) 135ζ(3) mχ3 1/4 − 3 ξ T (tdec) ΓB ηB . (63) After this the Universe is matter dominated and the en- ' 4 mχ3 2Γ− ergy density in radiation and χ3 redshift as The temperature just before decay can be arrived at by  8/3  2 1 teq 1 teq evolving the radiation energy density, resulting in ρrad = ρeq , ρχ3 = ρeq . (57) 2 t 2 t r  1/4 p 3 3 5 MPlΓχ Γ η 3 B . (64) We then assume that all of the χ3’s decay at the time B ' 8 π 2πg∗(Tdec) mχ3 2Γ− tdec = 1/Γχ3 . The ratio of the energy densities just before decay is Using the expression for the scattering rate in Eq. (52) + − evaluated at T (tdec), ρχ3 (tdec) −2/3 ξ − = (teqΓχ3 ) 3 r ≡ ρrad(t ) π πg (T ) Γ  dec η ∗ dec B  1/3   B g (T ) 50 ' 3ζ(3) 10 σscmχ3 Γχ3 MPl = 15 ∗ eq (58)  1/2 2   g∗(T0) g∗(T0) −11 g∗(Tdec)  mB  ΓB 9 10 −13 4/3  −22 2/3 ≈ × 50 5 GeV 10 GeV  mχ3  10 GeV . 8 GeV  10−22 GeV     × 10 GeV Γχ3 −5 . (65) − × mχ3 Γχ3 10 We use tdec here to indicate the time infinitesimally be- fore decay. Therefore we see that a baryon asymmetry of the required The dominance of the scattering rate over other scales size is possible for a heavy baryon system with  10−5, −2 ∼ in the problem allows us to make some simplifications which requires M12 /ΓB 10 with M12 / Γ12 not of the evolution equations that are useful here. In this small. | | ∼ | | | | limit we can ignore the Hubble rate as well as annihi- lation and the equations governing and ¯ in (47) can be integrated. This results in the evolutionB B equation for B. Full Solution of the Boltzmann Equations the difference between baryon and antibaryon densities, Eq. (48), becoming To get a more precise estimate of the baryon asymme-   try, we numerically solve the system in Eqs. (39), (40), d Γχ3 ρχ3 + 3H (nB nB¯ ) = and (47). As mentioned above, we need M12 /ΓB to −2 | | dt − mχ3 not be much smaller than around 10 . Looking at Ta- ∗ 0 2Im (M12Γ12) Brχ3→B ble I, one potential candidate is the Ωcb where the dom- 2 inant coupling involves the operator 2. In Fig. 5 × ΓB (ΓB + 2Γ−) + 4 M12 (dcb) | | Γ ρ Γ we show the value of ηB as a function of temperature in χ3 χ3 B 0 ¯ 0 , (59) the case of the asymmetry being sourced by the Ωcb-Ωcb ' mχ 2Γ− −12 3 system, taking mB = 7 GeV, ΓB = 3 10 GeV [34], −15 × ∗ which is valid for the cases we consider with M12 M12 = 3 10 GeV, Γ12/M12 = 0.3, arg(M12Γ12) = | |  | | × | | −23 ΓB Γ−. We have defined π/2, mχ3 = 7.5 GeV, Γχ3 = 3 10 GeV, and  × ∗ Brχ3→B = 0.35. We have used an annihilation cross sec- 2Im (M12Γ12) tion of 400 mb (the results do not depend on whether it  2 Brχ3→B ABBrχ3→B, (60) ≡ ΓB ' is flavor-blind or -sensitive) and the scattering rate given in Eq. (52). with AB from Eq. (22). Using , we can then relate the baryon asymmetry to In addition, the temperature dependence of the scat- tering and annihilation rates is compared to the expan- the χ3 number density at decay, sion rate of the Universe as well as to the rates governing −  − 3 nB nB¯ nχ3 (tdec) T (tdec) ΓB the baryon-antibaryon system in Fig. 6. As mentioned ηB = −+ = − +  before, the (decohering) scattering is the dominant pro- s(t ) s(t ) T (t ) 2Γ− dec dec dec (61) cess above temperatures of about 1 MeV and, in partic- −  − 3 3 T (tdec) T (tdec) ΓB ular, is always much larger than the annihilation rate. = ξ + . 4 mχ3 T (tdec) 2Γ− 13

�(�) At high temperatures, the heavy baryon density tracks ��-� ��-� ��-� ���� ��� its equilibrium value and it begins to deviate from its ��-� -� equilibrium value when χ3’s begin to decay. Although �� (�χ� /�) -� not directly evident from the plots (except through �� the change in the temperature vs. time), the out-of- equilibrium χ3 particles actually come to dominate the ��-�� ���� energy density of the Universe prior to their decay. Af- η� η� ter the χ3 decays, which we assume happens in less than ��-�� 0.1s, the Universe undergoes a transition from being ∼ _ matter-dominated to radiation-dominated, reheating to (�ℬ+�ℬ)/� ��-�� a temperature above a few MeV. We have numerically confirmed the rough accuracy of ��-�� the sudden decay approximation prediction for ηB over ��� �� much of the parameter space. Maximal CP violation, �(���) and thus more baryon asymmetry per oscillation, oc- ∗ curs for arg (M12Γ12) = π/2 and larger values of M12 FIG. 5. ηB = (nB − n ¯ )/s (solid, black) as a function of the | | B and Γ12 . A larger branching ratio, Brχ →B, would pro- temperature or time from a numerical solution of Eqs. (39), | | 3 0 ¯ 0 duce more oscillating baryons per Majorana decay. The (40), and (47) for parameters relevant to the Ωcb-Ωcb system: −12 −15 value of ηB that is generated is maximized if χ3 decays mB = 7 GeV, ΓB = 3 × 10 GeV, |M12| = 3 × 10 GeV, ∗ when the Universe’s temperature is about 10 MeV, i.e. |Γ12/M12| = 0.3, arg(M12Γ12) = π/2, mχ3 = 7.5 GeV, Γχ3 = −23 −2 3 × 10 GeV, and Br = 0.35. We have taken the τχ3 = 1/Γχ3 10 s. If it decays earlier than this, χ3→B ∼ rate for heavy baryon scattering on the plasma from Eq. (52) heavy baryon scattering on the plasma leads to decoher- and the annihilation cross section to be 400 mb. This can be ence, suppressing the asymmetry. If if decays later, the compared against the value of ηB (solid, gray) from a solution Universe does not have a sufficient baryon asymmetry of Eqs. (39), (40), and (59) as well as using the sudden decay at the time the begin to decouple, when the approximation (dashed, gray) in Eq. (65). Also shown are Universe is around 3 MeV. the ratio of the number density of χ3 to the entropy density −4 Given the constraints on the transition amplitudes in (multiplied by 10 , solid, orange) and the ratio of the B plus Table I, the most promising baryon that could allow for B¯ number densities to the entropy density (solid, purple). The dashed purple line shows the equilibrium B and B¯ density (in a large enough transition amplitude to source the BAU is the as yet unobserved Ω0 . A relatively large value units of the entropy density). The measured value of ηB = cb −11 for M12 is needed in this case, not far from the collider 8.8 × 10 is given by the solid red line. | | limit, unless Br 0 were rather large. It should be χ3→Ωcb noted that the collider limits discussed in Sec. V which appear in Table I depend on the specific model that we considered. It is conceivable that the model could be ex- �(�) tended in a way that makes the standard collider searches ��-� ��-� ��-� ���� ��� that we considered less constraining. For example, one could imagine making the φ decay to a large number ��-� Γ����� of relatively soft jets by coupling to a heavy vector-like quark and a singlet which decay to a large number of col- Γℬ ��-�� ored objects. Relaxing these limits could allow for other ��� heavy flavor baryons to source the BAU, potentially even observed baryons like the 0, 0, and 0. On the other -�� Ωc Λb Ξb ��� �� Γ�� hand, since they involve low-energy effective operators, 〈σ�〉(� the dinucleon decay constraints are less model depen- ℬ -�� +� _ �� ℬ ) dent. Weakening them would require significant tuning � of tree-level operators against those induced by weak in- teractions. ��-�� ��� �� �(���) VIII. SUMMARY AND OUTLOOK FIG. 6. The temperature dependence of the rates involved in the numerical solution of Eqs. (39), (40), and (47). The parameters are the same as in Fig. 5. In orange, from top We have presented a model for producing the observed to bottom are the scattering, annihilation, and Hubble rates. baryon asymmetry of the Universe which avoids high re- The purple lines indicate the rates relevant to the B-B¯ system heat temperatures. The asymmetry is generated through itself, ΓB, |M12|, and |Γ12|, from top to bottom, respectively. CP and B-violating oscillations of baryons occurring late in the hadronization era. Our model minimally intro- duces three neutral Majorana fermions and a single col- 14 ored scalar, and could potentially be embedded into RPV consequence of this scenario, and because of the require- SUSY. ment of baryon number violation involving heavy flavors, The Ωcb (scb) baryon emerges as our most promising it is likely to assume that dinucleon decay to kaons would candidate when∼ constraints due to collider data and din- be dominant. In the case of the LHC, a particular com- ucleon decay are taken into account. Note that the con- bination of signals in dijet resonances (singly and pair straints from colliders are more model-dependent than produced) along with an excess in jets plus missing en- those from the absence of dinucleon decay. Consider- ergy should be expected. We should mention in this case ing only the constraints from dinucleon decay, additional that a long-lived neutral particle, χ1, that decays hadron- 0 0 0 baryons, e.g., Ωc (ssc), Λb (udb), and Ξb (usb), ically is a generic prediction of this model. The typical ∼ ∼ ∼ 2−7 become viable candidates for baryogenesis via their os- χ1 decay length is in the range of 10 m, which could cillation. An interesting avenue for future work would be be well probed by the MATHUSLA detector that was re- constructing models that are less constrained by collider cently proposed. The signal of a long-lived but unstable experiments while preserving a large baryon oscillation particle at this experiment could help disentangle this rate. scenario from others that lead to excesses in jets plus Interesting signatures of this scenario could be present missing energy. in the large dataset of the upcoming Belle II experi- ment. If the lightest Majorana fermion is sufficiently light, one possible signature would be decays of heavy ACKNOWLEDGMENTS flavor hadrons that violate baryon number and involve missing energy. Additionally there could be heavy fla- We would like to thank Brian Batell, Kristian Hahn, vor baryons that oscillate into their at po- and Ahmed Ismail for useful conversations. The work of tentially measurable rates. Exploring the experimental DM is supported by PITT PACC through the Samuel prospects of this model at high luminosity, lower energy P. Langley Fellowship. The work of TN is supported colliders in more detail will be left for future work. by Spanish grants FPA2014-58183-P and SEV-2014-0398 Constraints from the LHC and the lack of dinucleon (MINECO), and PROMETEOII/2014/084 (Generalitat decay observation are quite important, suggesting the Valenciana). The work of AN and KA was supported in possibility of the detection of a signal in one or both part by the Department of Energy under grant number areas. Dinucleon decays are a more model-independent de-sc0011637.

[1] A. D. Sakharov, Pisma Zh. Eksp. Teor. Fiz. 5, 32 (1967), ph/0605297 [hep-ph]; M. Kawasaki, K. Kohri, T. Mo- [Usp. Fiz. Nauk161,61(1991)]. roi, and A. Yotsuyanagi, Phys. Rev. D78, 065011 (2008), [2] F. R. Klinkhamer and N. S. Manton, Phys. Rev. D30, arXiv:0804.3745 [hep-ph]. 2212 (1984); V. A. Kuzmin, V. A. Rubakov, and M. E. [5] P. W. Graham, D. E. Kaplan, and S. Rajendran, Phys. Shaposhnikov, Phys. Lett. B155, 36 (1985); P. B. Arnold Rev. Lett. 115, 221801 (2015), arXiv:1504.07551 [hep- and L. D. McLerran, Phys. Rev. D36, 581 (1987); Phys. ph]. Rev. D37, 1020 (1988). [6] L. Kofman, A. D. Linde, and A. A. Starobinsky, Phys. [3] M. S. Turner, F. Wilczek, and A. Zee, Phys. Lett. Rev. Lett. 73, 3195 (1994), arXiv:hep-th/9405187 [hep- B125, 35 (1983), [Erratum: Phys. Lett.B125,519(1983)]; th]; Y. Shtanov, J. H. Traschen, and R. H. Bran- D. Seckel and M. S. Turner, Phys. Rev. D32, 3178 (1985); denberger, Phys. Rev. D51, 5438 (1995), arXiv:hep- M. S. Turner and F. Wilczek, Phys. Rev. Lett. 66, 5 ph/9407247 [hep-ph]. (1991); P. Fox, A. Pierce, and S. D. Thomas, (2004), [7] S. Dimopoulos and L. J. Hall, Phys. Lett. B196, 135 arXiv:hep-th/0409059 [hep-th]; M. Beltran, J. Garcia- (1987). Bellido, and J. Lesgourgues, Phys. Rev. D75, 103507 [8] V. A. Kuzmin, (1996), arXiv:hep-ph/9609253 [hep-ph]. (2007), arXiv:hep-ph/0606107 [hep-ph]; D. J. E. Marsh, [9] D. McKeen and A. E. Nelson, Phys. Rev. D94, 076002 D. Grin, R. Hlozek, and P. G. Ferreira, Phys. Rev. Lett. (2016), arXiv:1512.05359 [hep-ph]. 113, 011801 (2014), arXiv:1403.4216 [astro-ph.CO]. [10] A. Ghalsasi, D. McKeen, and A. E. Nelson, Phys. Rev. [4] T. Moroi, H. Murayama, and M. Yamaguchi, Phys. D92, 076014 (2015), arXiv:1508.05392 [hep-ph]. Lett. B303, 289 (1993); M. Kawasaki and T. Mo- [11] S. Ipek, D. McKeen, and A. E. Nelson, Phys. Rev. D90, roi, Prog. Theor. Phys. 93, 879 (1995), arXiv:hep- 076005 (2014), arXiv:1407.8193 [hep-ph]. ph/9403364 [hep-ph]; M. Bolz, A. Brandenburg, and [12] M. I. Buchoff and M. Wagman, Phys. Rev. D93, 016005 W. Buchmuller, Nucl. Phys. B606, 518 (2001), [Erratum: (2016), arXiv:1506.00647 [hep-ph]. Nucl. Phys.B790,336(2008)], arXiv:hep-ph/0012052 [hep- [13] E. Friedman and A. Gal, Phys. Rev. D78, 016002 (2008), ph]; M. Kawasaki, K. Kohri, and T. Moroi, Phys. arXiv:0803.3696 [hep-ph]. Rev. D71, 083502 (2005), arXiv:astro-ph/0408426 [astro- [14] K. Abe et al. (Super-Kamiokande), Phys. Rev. D91, ph]; K. Kohri, T. Moroi, and A. Yotsuyanagi, 072006 (2015), arXiv:1109.4227 [hep-ex]. Phys. Rev. D73, 123511 (2006), arXiv:hep-ph/0507245 [15] M. Litos et al., Phys. Rev. Lett. 112, 131803 (2014). [hep-ph]; M. Kawasaki, F. Takahashi, and T. T. [16] J. L. Goity and M. Sher, Phys. Lett. B346, 69 (1995), Yanagida, Phys. Rev. D74, 043519 (2006), arXiv:hep- [Erratum: Phys. Lett.B385,500(1996)], arXiv:hep- 15

ph/9412208 [hep-ph]. 078, (2016), https://cds.cern.ch/record/2206252/ [17] H. Georgi, Weak Interactions and Modern Particle The- files/ATLAS-CONF-2016-078.pdf. ory (1984). [27] A. Monteux, JHEP 03, 216 (2016), arXiv:1601.03737 [18] A. G. Cohen, D. B. Kaplan, and A. E. Nelson, Phys. [hep-ph]. Lett. B412, 301 (1997), arXiv:hep-ph/9706275 [hep-ph]. [28] V. Khachatryan et al. (CMS), Eur. Phys. J. C75, 235 [19] C. Patrignani et al. (Particle Data Group), Chin. Phys. (2015), arXiv:1408.3583 [hep-ex]. C40, 100001 (2016). [29] M. E. McCracken et al., Phys. Rev. D92, 072002 (2015), [20] K. Jedamzik, Phys. Rev. D74, 103509 (2006), arXiv:hep- arXiv:1507.03859 [hep-ex]. ph/0604251 [hep-ph]. [30] M. Cirelli, P. Panci, G. Servant, and G. Zaharijas, JCAP [21] J. P. Chou, D. Curtin, and H. J. Lubatti, Phys. Lett. 1203, 015 (2012), arXiv:1110.3809 [hep-ph]. B767, 29 (2017), arXiv:1606.06298 [hep-ph]. [31] S. Tulin, H.-B. Yu, and K. M. Zurek, JCAP 1205, 013 [22] V. Khachatryan et al. (CMS), Phys. Lett. B747, 98 (2012), arXiv:1202.0283 [hep-ph]. (2015), arXiv:1412.7706 [hep-ex]. [32] S. Ipek and J. March-Russell, Phys. Rev. D93, 123528 [23] ATLAS Collaboration Public Note (ATLAS), ATLAS- (2016), arXiv:1604.00009 [hep-ph]. CONF-2016-084, (2016), https://cds.cern.ch/ [33] M. H. Thoma, (1995), arXiv:hep-ph/9503400 [hep-ph]. record/2206277/files/ATLAS-CONF-2016-084.pdf. [34] V. V. Kiselev and A. K. Likhoded, Phys. Usp. 45, [24] M. Aaboud et al. (ATLAS), Eur. Phys. J. C76, 547 455 (2002), [Usp. Fiz. Nauk172,497(2002)], arXiv:hep- (2016), arXiv:1606.08772 [hep-ex]. ph/0103169 [hep-ph]; Z. Shah, K. Thakkar, and A. K. [25] CMS Collaboration Public Note, CMS-PAS-SUS-16-014, Rai, Eur. Phys. J. C76, 530 (2016), arXiv:1609.03030 (2016), https://cds.cern.ch/record/2205158/files/ [hep-ph]; W. Wang, F.-S. Yu, and Z.-X. Zhao, (2017), SUS-16-014-pas.pdf. arXiv:1707.02834 [hep-ph]. [26] ATLAS Collaboration Public Note, ATLAS-CONF-2016-