NMSSM with a singlino LSP: possible challenges for searches for supersymmetry at the LHC

A light singlino in the NMSSM can reduce considerably the missing transverse energy at the end of sparticle decay cascades; instead, light NMSSM-specific Higgs bosons can be produced. Such scenarios can be consistent with present constraints from the LHC with all sparticle masses below ~1 TeV. We discuss search strategies, which do not rely on missing transverse energy, for such scenarios at the next run of the LHC near 14 TeV.


Introduction
One of the main tasks of the LHC was and will be the search for supersymmetric (SUSY) particles. The largest production cross sections are expected for gluinos (g) and squarks (q) of the first generation. After the first run of the LHC at a center of mass (c.m.) energy of mostly 8 TeV, no significant excesses have been observed in corresponding search channels [1,2] (see [3] for a recent summary).
The absence of excess events can be interpreted in terms of lower bounds on gluino and squark masses, once assumptions on their decay cascades are made. These depend on the masses and couplings of many other SUSY particles, at least on the mass of the lightest SUSY particle (LSP). Within simplified models (assuming simple 1-step decay cascades) or the Minimal SUSY extension of the Standard Model (MSSM), lower bounds on gluino and squark masses are typically in the 1.2-2 TeV range, and ∼ 1.7 TeV if gluino and squark masses are assumed to be similar [1,2]. Although these constraints do not rule out the MSSM, they eliminate a significant part of its "natural" parameter space [4].
However, the MSSM is not the only SUSY extension of the Standard Model (SM) which alleviates the hierarchy problem, provides an acceptable dark matter candidate and leads to Grand Unification of the running gauge couplings. In the present paper we consider the Next-to-Minimal SUSY extension of the Standard Model (NMSSM) [5], where the coupling of the two Higgs doublets of the MSSM to an additional gauge singlet field S renders more natural a value of ∼ 126 GeV of the SM-like Higgs boson [6][7][8][9][10][11], while preserving the attractive features of the MSSM. Besides the Higgs sector, the NMSSM differs from the MSSM through the presence of an additional neutralino (the singlino, the fermionic component of the singlet superfield). The singlino can be the LSP, which can modify considerably the SUSY particle decay cascades [12][13][14][15][16][17][18][19][20].
The strongest constraints from searches for gluinos and squarks of the first generation originate from channels where one looks for events with jets with large transverse momentum p T and missing transverse energy E miss T [1,2]. The E miss T is due to having all SUSY decay cascades ending in a stable LSP (under the assumption of R-parity conservation), which escapes detection (if neutral, as required for dark matter).
In the present paper we point out that a singlino-like LSP in the NMSSM can reduce significantly the missing transverse energy at the end of SUSY particle decay cascades. This is due to the kinematics of the last process in a SUSY particle decay chain, NLSP → LSP + X, where NLSP denotes the Next-to-LSP, and X a particle (e.g. a Higgs boson) decaying into visible components of the SM. For a light LSP, if the mass of X is close to the mass of the NLSP, little energy and momentum are transferred from the NLSP to the LSP; most of the energy is transferred to X. Correspondingly, the LSP in the final state leads to little E miss T , whereas large E miss T is one of the relevant search criteria for SUSY particles in general.
The possibility to discover squarks and gluinos without relying on E miss T , but on leptons, has been studied earlier in [21][22][23]. [23] discuss decays of an NLSP into a scalar (decaying visibly into SM particles) and the LSP, referring to the NMSSM without, however, considering the particular kinematic configurations analysed below.
A scenario similar to the one discussed here has been named "Stealth Supersymmetry" [24,25]. There, however, a complete "stealth sector" is added to the MSSM in order to obtain the above kinematic configuration of the NLSP decay.
An extensive survey of present constraints on gluinos from searches, including several without relying on E miss T , is given in [26]. Among the scenarios analysed in [26] are socalled "minimal Hidden Valley" models. These are similar to the ones considered here after replacing the extra singlet scalar and its fermionic superpartner [26] by the corresponding states of the NMSSM (and the NLSP higgsino by a bino-like NLSP). It was already found in [26] that the kinematic configuration discussed above leads to the weakest constraints.
The present scenario is opposite to the one of "compressed supersymmetry" [27][28][29][30][31][32] where the masses of the NLSP and the LSP are assumed to be similar, and little energy is transferred to X. Then jets (or leptons) with large p T would be rare. Moreover, unless a hard jet is emitted from the initial state ("monojet"), the E miss T due to two LSPs emitted from two SUSY particles back-to-back in the transverse plane tend to cancel.
In the MSSM, the kinematic configuration considered here cannot play a major rôle: A light LSP (with a mass of a few GeV) can only be bino-like, since winos or higgsinos would have charged partners with similar masses, already ruled out by LEP. All squarksappearing also in gluino decays -have hypercharge and hence couple to the bino. If the LSP is a very light bino, squarks will in general decay directly into the bino, and hardly pass through an NLSP (e.g. a heavier neutralino or chargino) and a state X (a Higgs, Z or W boson). Thus only a fraction of cascade decays leads to a reduction of E miss T , so that the interpretation of the absence of signal events in terms of lower bounds on SUSY particle masses remains practically unchanged.
On the other hand, in the NMSSM the bino can be the NLSP, the singlino a light LSP, and X a priori a Higgs, a Z or even a W boson (if the NLSP is a chargino). Then the decays of X can still give rise to missing energy in the form of neutrinos; this is the case for the decays of the W and Z bosons, and also for the SM-like Higgs (when it decays via W W * or ZZ * ).
However, in the NMSSM additional Higgs bosons exist, which can be lighter than the Z boson and are not excluded by LEP due to small couplings to ZZ. A lighter CP-even Higgs boson H 1 with a mass below M Z would have very small decay rates into W W * , but decay dominantly into bb and, to some extent, into τ + τ − . Although the latter decays can also give rise to some E miss T , the scenario NLSP → LSP + H 1 with M H 1 < ∼ M NLSP < M Z would be the most difficult one with respect to signatures based on E miss T . (Subsequently we denote scenarios with as little E miss T as possible as "worst case".) In the case of squark/gluino pair production, some E miss T can also originate from W , Z and/or Higgs decays which appear during decay cascades involving charginos and/or heavier neutralinos. Again, a "worst case" scenario would be one where this does not happen if, for instance, the chargino and heavier neutralino masses are close to (or above) the squark masses.
In the present paper we concentrate on such "worst case" scenarios: First, we present the properties of points in the NMSSM which are not excluded by present SUSY searches although all sparticle masses are below ∼ 1 TeV. Second, we propose search strategies for these difficult scenarios, putting forward an analysis of events at the LHC near 14 TeV c.m. energy, based on the decay products of two H 1 bosons in the bbτ + τ − + jets final state. Our simulations indicate that, for not excessively heavy squarks and gluinos (i.e. a not too small production cross section), a signal can be visible above Standard Model backgrounds.
In the next section we discuss in detail scenarios within the general NMSSM, in which E miss T is reduced for kinematic reasons. Results of event simulations of such a benchmark point are discussed, which explain the reduced sensitivity of present SUSY searches to such a scenario. We also discuss simplified models with varying LSP and H 1 masses, and the corresponding reduction of signal events. In Section 3 we attempt to extract signals for H 1 pair production at the LHC with 14 TeV c.m. energy, with dedicated cuts which do not rely on E miss T . Instead, we attempt to identify b-jets and τ -leptons from boosted H 1 bosons with the help of a jet reconstruction with a small jet cone radius R = 0.15. Section 4 contains a summary and conclusions.
2 "Missing" missing energy in the NMSSM Given a possible last step in a SUSY particle decay chain NLSP → LSP + X in the limit of a narrow phase space, M NLSP − (M LSP + M X ) ≪ M NLSP , the energy (momentum) transferred from the NLSP to the LSP in the laboratory frame is proportional to the ratio of masses: Hence, if the LSP is light, little (missing transverse) energy is transferred to the LSP; the transverse energy is carried away by X. The effect is the more important the narrower the phase space is. As explained in the Introduction, such a scenario is difficult to realise in the MSSM where such a light LSP must be bino-like. The particle content of the NMSSM differs from that of the MSSM by an additional singlino-like neutralinoS, and additional singlet-like CP-even and CP-odd Higgs bosons [5]. Notably the NMSSM spectrum contains three CP-even Higgs bosons H i , i = 1, 2, 3 (ordered in mass). The singlino-like neutralino can be the LSP with a bino-like NLSP (as occurs for the regions of parameter space considered below). Then the above scenario of little E miss T being transferred to the LSP can be realised with a singlet-like CP-even Higgs boson H 1 playing the rôle of X, whose subsequent decays give rise to little invisible transverse energy in the form of neutrinos. Typical values for the masses would be a few GeV for the singlino-like neutralinoS, a bino-like NLSP with a mass M bino just below M Z , and M H 1 just below M bino − MS. Note that, due to its reduced coupling to the Z boson, such a light H 1 can still be compatible with constraints from Higgs searches at LEP [33].
In the simplest Z 3 invariant realisation of the NMSSM, the diagonal elements of the mass matrices for the (pure) singlet-like statesS, H S and A S satisfy [19] [34]. Hence we consider in the following a general NMSSM, still with a Z 3 invariant superpotential In the above hatted letters denote superfields, and the ellipses denote the MSSM-like Yukawa couplings ofĤ u andĤ d to the quark and lepton superfields. We allow for the following NMSSM specific soft SUSY breaking terms As can be seen from Eq. (2.3), a vacuum expectation value S generates an effective µ eff term µ eff = λ S , which has to be larger than ∼ 100 GeV for the charged higgsinos to satisfy bounds from LEP. Given the diagonal singlino mass term MS = 2κ S , a singlino mass of a few GeV is obtained for κ about two orders of magnitude smaller than λ.
For completeness we comment on the possibilities to obtain consistent properties of dark matter in such a scenario. Within GMSB, a gravitino can be lighter thanS which would thus not be the "true" LSP, but decay radiatively into a gravitino and a photon (through a small photino component from a non-vanishing mixing with the bino/wino). However, the singlino life time would be so large that this decay would occur outside the detectors and have no impact on our subsequent analyses. (On the other hand, the singlino life time should not exceed ∼ 100 s in order not to spoil nucleosynthesis unless the NLSP density is diluted through entropy production.) Alternatively, the singlino relic density can be reduced to comply with the observed dark matter relic density through the exchange of a CP-odd Higgs state A S in the s-channel, provided M A S ∼ 2MS. We have checked that this is indeed possible, and the benchmark point given below has this property.
Returning to the issue of E miss T , its suppression is maximised if no neutrinos from Z/W decays are emitted during squark/gluino decay cascades. In a truly "worst case scenario" winos, higgsinos, sleptons, stops and sbottoms are not produced neither in squark nor in gluino decays. Bino → Z+ singlino decays are impossible, if the bino mass is below M Z . In Table 1 we give the parameters and particle masses of a benchmark point with these properties, for which physical masses and decay branching fractions have been obtained with the public code NMSSMTools 4.2.1 [35,36]. Table 1 denote the soft SUSY breaking bino-, wino-and gluino mass terms and Higgs-stop, Higgs-sbottom trilinear couplings, respectively. The lightest neutralinos are denoted by χ 0 1 (singlino-like) and χ 0 2 (bino-like), respectively. The gaugino mass terms are non-universal, but lead to a go-theorem for branching fractions corresponding to a simplified model: Squarks decay to 100% into χ 0 2 and the corresponding quark, χ 0 2 to 100% into χ 0 1 + H 1 . Gluinos decay with approximately equal branching fractions only into squarks + quarks of the first two generations. Hence the decay chains areq Table 1: Parameters (left and middle column) and particle masses (right column) of a NMSSM benchmark point.
The small value of λ suppresses mixings of the singlet-like A 1 with the doublet-like A 2 such that χ 0 2 decays into χ 0 1 + H 1 in spite of the lighter A 1 , and suppresses mixings of the singlet-like χ 0 1 with higgsinos/winos which can lead to unacceptable invisible decay rates of the Standard Model-like H 2 into χ 0 1 + χ 0 1 . Despite the small value of λ (but due to the large value of tan β) the mass of H 2 -still consistent with ∼ 126 GeV within the expected theoretical error in NMSSMTools -is larger than in the MSSM due to mixing of H 2 with H 1 [37]. All phenomenological constraints (except for the muon anomalous magnetic moment) tested in NMSSMTools are satisfied. Due to the small width of A 1 it is difficult, however, to determine the dark matter relic density accurately with the code micrOMEGAs [38] inside NMSSMTools -for our benchmark point its numerical value seems smaller than the desired value Ωh 2 ∼ 0.12 [39,40], which shows in any case that the relic density can be reduced sufficiently.
We have simulated events at the LHC at 8 TeV for this point using MadGraph/-MadEvent [41] which includes Pythia 6.4 [42] for showering and hadronisation. The emission of one additional hard jet was allowed in the simulation; the production cross sections for squark-squark, squark-gluino, squark-antisquark and gluino-gluino production were obtained by Prospino at NLO [43,44], including correction factors from the resummation of soft gluon emmission estimated from [45,46]. (At 8 TeV, the total squark-gluino production cross section for the benchmark point is ∼524 fb.) The output in StdHEP format was given to CheckMATE [47] which includes the detector simulation DELPHES [48] and compares the signal rates to constraints in various search channels of ATLAS and CMS. Additional analyses were performed by means of MadAnalysis 5 [49,50].
Following CheckMATE, the signal rates for the above benchmark point are compatible with constraints from available search channels for SUSY. In spite of the many b-jets from H 1 decays, the dominant constraints for this point originate from the search for jets and E miss Searches for events at the LHC with jets and bb pairs have also been performed by ATLAS in [51] however aiming at resonances in the bbbb final state. Also the upper bounds on signal rates in bbγγ final states from CMS [52] and ATLAS [53] are satisfied, amongst others since the γγ invariant mass required there does not cover our range of M H 1 .
An M T 2 Higgs analysis was performed by CMS in [54], which aimed at a scenario similar to those discussed here: squark/gluino decay cascades ending in χ 0 2 → χ 0 1 + H SM . In one of the considered channels (high H T region) the cut on E miss T was lowered to E miss T > 30 GeV. The absence of significant excesses was interpreted in terms of a gluino-induced simplified model leading to M gluino > ∼ 850 GeV (depending in M χ 0 1 ); however, M χ 0 2 = M χ 0 1 + 200 GeV was assumed in this analysis.
Further searches for excesses in events with jets without lower cuts on E miss T have been performed in [55][56][57][58]. The most constraining search channel in [55] is the one requiring 7 jets with p T > 80 GeV and at least two tagged b-jets. After a simulation we find about 240 events in this channel for the benchmark point, complying with the data at the 2 σ level. Concerning the search for three-jet resonances in [56], we find that the two b-jets from H 1 decays merge sufficiently often into one single jet such that the event rates and acceptance are about 20 times smaller than the one assumed in [56] for gluino pair production with RPV decays into three jets, and the limits are well satisfied. Regarding the search for twojet resonances in [57] we find that the average two-jet mass peaks at ∼ 800 GeV (somewhat below the squark/gluino masses) and the acceptance after cuts (within a window of a width of ∼ 15% times the mass) is < ∼ 4%; consequently the signal complies with the limits shown in Fig. 3 in [57]. Finally we have studied S T , the sum of |p T | of objects with |p T | > 50 GeV relevant for the search for microscopic black holes in [58]. For all multiplicities N > ∼ 3...10 the signal events are below 10% of the data points shown in Figs. 2 and Figs. 3 in [58] without any peak-like structure, and thus this search is not restrictive. Hence present analyses, potentially sensitive to the decay products of two H 1 bosons in the final state, are not sensitive to the benchmark point.
Given that the masses of the gluino and the squarks of the first generation are well below 1 TeV it is clear that a corresponding point in the parameter space of the MSSM would be well excluded, the reason being the different spectra of E miss T . To clarify this effect we show in Fig. 1 the spectrum of E miss T for the benchmark point and for a similar point in the MSSM, which differs from the benchmark point only in a stable bino, which is now the LSP.
In Fig. 1 one sees the dramatic reduction of E miss T due to the NLSP → LSP + H 1 decay; the few remaining events with large E miss T for the benchmark point (denoted by NMSSM BMP in Fig. 1) originate from neutrinos from τ and b decays after H 1 → τ + τ − , bb (and, to a minor extent, from the singlino).
In fact, the final states from H 1 decays not only reduce E miss T , but lead also to an increase of m eff (Nj). Hence the cut E miss     Table 2 has a stronger increase with decreasing M H 1 (away from the boundary of phase space).
As stated above, the impact of the "missing" E miss We see that the reduction of signal events remains very strong, even if M H 1 is several GeV away from the boundary of phase space. Hence the result of the analyses summarised in Table 1 and Fig. 2 is that suppressing the number of signal events in typical SUSY search channels does not require a particular fine-tuning of masses.
On the other hand we should keep in mind that contributions from neutrinos from squark decay cascades to E miss T have been suppressed for all points considered above; still it seems important to re-interpret the absence of excesses in typical SUSY search channels in terms of lower bounds on squark/gluino masses for such configurations of bino-, singlinoand H 1 masses.
The next question is whether other search strategies, not relying on strong cuts on E miss T , can be sensitive to such difficult SUSY scenarios with two H 1 bosons in the final state. A first proposal towards extracting corresponding signals -which will be improved in a future publication -is presented in the next section.
3 Towards the extraction of signals in bb + τ + τ − final states at 14 TeV Signals from Higgs production via neutralino decays in sparticle decay cascades have been analysed previously, mostly in the context of the MSSM in [17,[59][60][61][62][63][64][65][66][67][68]. There, however, significant lower cuts on E miss T were applied, since in the considered scenarios the LSP had no reason to be particularly soft.
In the present scenario the final states of squark and gluino production are characterised by jets with large p T , little E miss T , but remnants of two H 1 Higgs bosons whose branching ratios coincide essentially with those of a Standard Model-like Higgs boson of the corresponding mass. (The prospects for a direct discovery of H 1 at the LHC are rather dim: its couplings to SM particles squared, and hence its production cross sections and signal rates, are only 6% − 6.5% of the ones of a SM-like Higgs boson of similar mass.) Subsequently we describe an approach towards the extraction of a possible signal in the bb + τ + τ − final state, using the same simulation methods described in the previous section.
First we have studied E miss T for the benchmark point for pp collisions at the LHC at 13 TeV and 14 TeV c.m. energy assuming an integrated luminosity of 100 fb −1 . As can be seen in Fig. 3, E miss T is still peaked at low values.  When looking for the remnants of two H 1 Higgs bosons, an important question is which transverse momenta can one expect for these particles. In Fig. 4 we show the leading and next-to-leading p T distribution corresponding to the benchmark point. (Here and in the following we concentrate on 14 TeV c.m. energy.) We see that the p T of the leading H 1 is peaked near 400 GeV, and the p T of the next-to-leading H 1 is peaked near 200 GeV, which can be used for cuts on the final states.
We next describe the sequence of cuts which were applied. The analysis of the events was performed in two steps: to start with, jets were constructed by Fastjet [69] (part of the Delphes package [48]) using the anti-k T algorithm [70] and a jet cone radius R = 0.5. This value was chosen such that, as often as possible, all decay products of the leading H 1 (but no other hadrons) are part of a single jet whose mass distribution is analysed at the end.
We require four hard jets (including b-tagged jets) with p T > 400 GeV, 200 GeV, 80 GeV, 80 GeV, respectively. A significant E miss T is not part of the signal; some E miss T can be expected, however, from neutrinos of τ decays once we require 2 τ in the final state (see Fig. 3). Hence we only impose E miss T > 30 GeV. Then, jets of the same event were reconstructed with a jet cone radius R = 0.15. The aim is to identify as many "slim" b-jets as possible, together with their kinematic properties. The value R = 0.15 is just marginally larger than the granularity 0.1 × 0.1 of the ATLAS hadronic tile calorimeter (we use the ATLAS detector card inside Delphes). For b-tagged jets we require p T > 40 GeV and assume a b-tag efficiency of 70% (mistag efficiencies from c-jets of 10%, and from light quark/gluon jets of 1%).
Among the jets reconstructed with R = 0.15 we require ≥ 2 b-jets and ≥ 2 hadronic τ leptons. Since the invariant mass of the pair of τ -leptons is difficult to reconstruct we just require M τ τ < 120 GeV and, in order to remove the background from fake τ leptons (see below), M τ τ > 20 GeV. The 2 b-jets next to each other (with the smallest ∆R) are combined into a 2b-pseudojet, 2bPJ. Among the jets constructed with a jet cone radius R = 0.5 we look for the one closest to the 2bPJ (with ∆R < 0.1); this jet J is our candidate for the remnants of H 1 → bb. For J we further require p T > 400 GeV. The invariant mass of the 2bPJ should be smaller than the mass of J . (The mass of the 2bPJ can be considerably smaller due to radiation off the b-quarks not included in the R = 0.15 jets.) Finally we require the mass of J to be 40 GeV < M J < 120 GeV, and plot M J in this range.
The result displayed in Fig. 5, which is based on the simulation of ∼ 230000 events, shows that M J peaks indeed near the mass of H 1 , 83 GeV for the benchmark point considered here. The event rates are normalised to an integrated luminosity of 100 fb −1 ; for the total squarkgluino production cross section at 14 TeV we have 5232 fb at NLO+NNLO. The impact of the above cuts is shown in Table 3; within the signal region 40 GeV< M J < 120 GeV the cross section is about 23 fb. Given the H 1 → τ + τ − branching fraction of about 8% and the tagging efficiencies, the dominant reduction of signal events results from the requested ≥ 2 b-jets and ≥ 2 τ leptons.
A priori, the following Standard Model backgrounds contribute to the final states de-  Table 3: Impact of the cuts described in the text on the event rates of the benchmark point and the dominant jjbb background (the latter after cuts at the parton level as described in the text).  Table 3. The background contribution from jjbb with two mistagged τ leptons is shown in blue.
fined above: jjbbτ + τ − from QCD and electroweak production, jjtt possibly with additional mistagged τ leptons, and jjbb with two mistagged τ leptons. We have estimated these backgrounds using the same simulation methods applied for the signal. However, in order to make our cut analysis sufficiently efficient, we also applied cuts at the parton level in MadGraph on jets (quarks and gluons) and b-quarks: For b-quarks we required p T > 100 GeV and |η| < 2, and for the four leading jets (including b-quarks) p T > 200 GeV, 100 GeV, 80 GeV, 80 GeV, respectively. We checked that these cuts do not generate a bias in the M J spectrum after applying the additional cuts described above and in Table 3.
After the cuts, the background from jjbbτ + τ − from QCD and electroweak production turned out to be negligibly small, with a cross section in the signal region of about 0.007 fb. After all cuts, the background from jjtt is also small, with a cross section in the signal region of about 0.44 fb.
However, the background from jjbb with two mistagged τ leptons is relatively large due to the fact that we tag for b-jets and τ -leptons using jet reconstruction with a small jet cone radius R = 0.15. Many of such "slim" jets are mistagged as τ -leptons, often as pairs with a relatively low invariant mass. A priori, about 5% of all jjbb events after cuts contain such a fake τ pair. This fake rate can be reduced by a factor ∼ 1 2 after a cut M τ τ > 20 GeV, which reduces the signal by only about 12%. Then this background results in a cross section in the signal region of about 7 fb. Its contribution to M J , based on the simulation of 300000 events, is also shown in Fig. 5, and it seems that the signal can be distinguished clearly from this background.
In practise the background is often obtained from data-driven control regions. Once it is measured, modifications of the cuts given above and/or additional cuts (for example on the absence of isolated leptons) are likely to improve the signal/background ratio even further.
Of course, the mass M H 1 can differ from the value M H 1 = 83 GeV assumed for the benchmark point. In order to see the impact of a lighter H 1 we have repeated the simulation for a point with the same sparticle spectrum but M H 1 = 60 GeV, M NLSP = 67 GeV and M LSP = 5 GeV. The resulting M J spectrum is shown in Fig. 6 and again we see that the sequence of cuts allows, in principle, to identify M H 1 from the M J spectrum.
The H 1 H 1 final state, characteristic of squark/gluino production in the present scenario, resembles actually Higgs pair production in the Standard Model up to the unknown H 1 mass but, provided that squarks and gluinos are not excessively heavy, with an associated larger cross section (see [71][72][73][74][75] for some recent studies in the Standard Model). Corresponding techniques like more refined subjet-based analyses applied to the bb τ + τ − final state in [71] can probably be useful here as well.
The squark/gluino production cross sections would decrease, of course, for squarks  Table 3. and/or gluinos heavier than their benchmark value (860/890 GeV, respectively), and the kinematics will change. First we investigate the impact of heavier squarks and gluinos on the shape of E miss T . We illustrate this with simplified models where, as in the case of the previous benchmark point, squarks decay with a 100% branching ratio into the bino-like NLSP (still with a mass of 89 GeV) which can only decay (with 100% BR) into H 1 and the singlino-like LSP (both still with masses of 83 GeV and 5 GeV, respectively). The corresponding E miss T distributions are shown in Fig. 7 for squark/gluino masses of 1000 GeV or 1400 GeV. (The gluinos are taken 5 GeV heavier than squarks to allow, for simplicity, for flavour democratic gluino 3-body decays into quarks + squarks of the first two generations.) One finds that E miss T still strongly peakes at low values; hard cuts on E miss T would remove again most of the signal events. The transverse momenta of the the leading and next-toleading H 1 for squarks/gluinos with masses of 1000 GeV and 1400 GeV, respectively, are shown in Fig. 8.
As visible from Fig. 8, the average transverse momenta of the H 1 states are considerably larger for heavier squarks/gluinos. Finally we ask whether the shape of the M J spectrum changes for heavier squarks/gluinos. Using the same analysis and cuts as before, the resulting M J spectrum is shown in Fig. 9 for squark/gluino masses of 1000 GeV, 1200 GeV and 1400 GeV. We see that the shape of the M J spectrum remains unchanged; only the signal rate decreases as expected (slightly less, in fact) as does the production cross section which is now about 2226 fb, 693 fb and 242 fb for squark/gluino masses of 1000 GeV, 1200 GeV and 1400 GeV, respectively. However, for squark/gluino masses above 1400 GeV, the signal obtained with the present cuts (and jet analysis) starts to fall below the background from jjbb with two mistagged τ leptons (shown in Fig. 5).
On the other hand, since the production of heavier squarks and gluinos generates both H 1 states and jets with larger transverse momenta, cuts can be optimised. Search channels with significantly harder cuts on the transverse momenta of at least the candidate J-jets (assumed to contain the remnants of the leading H 1 ) can be employed. Corresponding   analyses will be the subject of future publications.

Conclusions and outlook
The most important result of the present paper is the existence of scenarios in the general NMSSM in which a light singlino at the end of sparticle decay cascades reduces strongly the missing transverse energy, one of the essential criteria in standard search channels for supersymmetry. In such "worst case scenarios" hardly any missing transverse energy is produced along each step of sparticle decay cascades. We present a realistic benchmark point, consistent with the properties of the Standard Model-like Higgs boson at ∼ 126 GeV and the dark matter relic density, satisfying present constraints from SUSY search channels with all sparticle masses below ∼ 1 TeV. The two NMSSM-like Higgs bosons H 1 , produced in each event of squark, squark-gluino or gluino pair production in this scenario, allow for new search channels which do not rely on large missing transverse energy, but on the H 1 decay products. We have presented an analysis which shows that, for not too heavy squarks/gluinos, a H 1 signal can be visible above the Standard Model background in the bbτ + τ − + jets final state. This analysis can certainly be improved in various aspects, but already indicates the lines along which a signal can be obtained.
Among the possible improvements are analyses based on jet substructure as is the case of Higgs pair production into the same final state in [71] (replacing the step of our analysis based on a jet cone radius of 0.15), which may also help to reduce the background from mistagged tau pairs. Also searches for a (bb)(bb) final state as in [75] might be feasible. Finally, in order to get some direct information on the masses of the originally produced squark/gluino pairs (beyond the production cross section), analyses based on jet substructure may be combined with analyses based on M T 2 as, for example, in [54].
Variants of the benchmark scenarios discussed here could also be realised in principle: First, winos and/or higgsinos could be lighter and appear in squark decay cascades. Then standard SUSY search channels relying on E miss T and jets, isolated leptons etc., start to become sensitive to squark/gluino production but it remains to be studied when, in the presence of a light singlino and for the kinematical situation discussed here, such search channels become more sensitive than the type of analysis presented here.
Second, the final state "X" in the final step NLSP → X + LSP of sparticle decay cascades does not necessarily have to be a light NMSSM-specific Higgs boson H 1 . Again, if X is for instance a Z boson or a combination of Z and H SM bosons (depending on the branching fractions of the NLSP), standard SUSY search channels can become relevant since more E miss T is expected from X decays. However, if the number of events with E miss T is still reduced due the particular kinematical situation discussed here, channels which depend less on E miss T but more on X decay products would again be more promising. Such cases merit also to be studied in the future.