In recent publications [1–3], we have presented experimental indications of the possible existence of a new at least ternary decay channel of low excited heavy nuclei known as collinear cluster tri-partition (CCT). A fragment mass M is calculated by the energy E and the velocity V. Mainly a scattering of fragments at the entrance of an E-detector gives background events simulating ternary decay. Selection of the “true” events was provided by applying the gates on the fragments momenta, velocities, experimental neutron multiplicity, and the parameters sensitive to the fragment nuclear charge. Observation of the specific linear structures in the M –M distributions (mass correlation plots) served as a criterion for a sufficient suppression of the background. The structures were reproduced at the spectrometers of two types. Earlier experiments were performed using gas filled detectors (modules of the FOBOS setup ). Later we switched to solid-state detectors, namely timing detectors, on the microchannel plates and the mosaics of PIN diodes (COMETA setup  and the similar ones ). Even though mass reconstruction procedures for these two types of spectrometers strongly differ, the obtained results are in good agreement.
Estimation of the expected parameters of the CCT products was performed in the recent theoretical works [5–7]. The results obtained were taken into account in our model of the most populated CCT mode known as “Ni-bump”.
2. Experiments and results
Figure 1(a) shows the region of the mass distribution measured at the COMETA setup  in the experiment Ex1 for the fission fragments (FFs) from Cf (sf) around the Ni-bump (M = 68–80 amu, M = 128 150 amu). The structures are seen in the spectrometer arm facing the source backing only. No additional selection of the fission events was applied in this case, which resulted in the experiment having almost no background. A rectangular-like structure below the locus of binary fission is bound by magic nuclei (their masses are marked by the numbered arrows), namely Sn (1), Ni (2), and Ni (3). In Figure 1(b), we demonstrate the projection of the linear structure seen at the masses of 68 and 72 amu.
Similar structures were revealed as well in the experiment Ex2 performed at the COMETA-F spectrometer which differs substantially from this used in Ex1 by the data acquisition system based on the fast flash-ADC (CAEN DT5742) and data processing .
In fact, only two fragments were detected in each decay event. The mass and velocity of the “missed” fragment could be calculated based on the laws of mass and momentum conservation. In each event showing the missing mass (ternary event), we mark the masses of the fragments in order of their decreasing masses M , M and M (Ternary particle) respectively. Figure 2(a) demonstrates a correlation between the velocities of two lighter partners of the ternary decay. Only the events for which M = (67–75) amu (Ni-peaks in Figure 1(b)) are under analysis. Their total yield does not exceed 2.5x10 per binary fission. Three different groups of events are vividly seen in the figure. They are marked by the signs w1–w3 respectively. Each of the loci consists of two subgroups as can be inferred from the plot E –E (Figure 2(d)).
3. Scission point calculations
Among all the theoretical articles initiated by the results of the experiments, article  deserves special attention. Under the three-center shell model the potential energy surfaces for few ternary combinations in a fission channel were calculated for the Cf nucleus. The fission barrier for the Sn + Ca + Ni ternary splitting is shown in Figure 3(a). According to the figure, the exit point corresponds to R 22.4 fm, i.e. elongation of the system exceeds the length of the configuration of three touching spheroids. If just a Ca nucleus took upon itself all extra elongation, the axis ratio of the corresponding spheroid would be approximately 1:1.6.
Calculations  performed in ten dimensional deformation space demonstrate the shapes of a decaying Cf nucleus at large deformations (Figure 3(b)) in the potential valleys 3 and 4. The distance between the centers of the side constituents (R ) are equal to approximately 18 fm and 23 fm. After the rupture at the narrowest section of the neck, almost all deformation energy concentrates in the light (panel c) or heavy fragment (panel d).
Typical shapes of a fissioning nucleus at large deformations are confirmed independently by the neutron data from . For the Cm, such asymmetry in number of emitted neutrons from the light (ν ) and heavy (ν ) FFs was traced up to the ν /ν = 9/0. Just for a sense of the scale of yields of highly deformed scission configurations, one can cite to the relative total yield Y of the fission events at ν = 6 and ν / ν = 6/0. Y was estimated to be 2.19 % and 0.72 % for Cm and Cf respectively .
The following scenario of the CCT process can be proposed based on our experimental findings and recent theoretical calculations. According to Ref. , the exit point from under the barrier in the potential valley leading to the Sn + Ca + Ni ternary splitting (Figure 3(a)) corresponds to a much more elongated configuration in comparison with the chain of three touching spherical nuclei. The distance R between the centers of the side clusters was estimated to be above 23 fm. Likely, the central fragment (Ca) takes upon itself almost all extra elongation. After a rupture occurs, for instance on the boundary of Ca and Sn clusters, the Ni cluster very quickly (in comparison to full acceleration time) attracts the Ca “neck”. Part of the released deformation energy is spent on emission of neutrons flying apart isotropically. Thus, the formed pear-shaped Ni-Ca dinuclear system can rotate around the center of its gravity by 180 . Such orientation is the most energetically favorable. Octupole vibrations could be another reason for the change in the orientation of the “tip”. Formation of the Ca-Sn system with similar features is less probable . The formed dinuclear system can evolve towards fusion or rupture. In the first case, we deal with a binary fission of a mother nucleus, and in the second instance with a ternary fission.
Presumable decay scenarios for all subgroups are presented in Table I. A precission configuration of the system is demonstrated in the third column of the table. For all the cases fission fragment FF is supposed to be Ni, the mass of the FF corresponds to the mean mass of the lightest cluster (shown in brackets in Figure 3(d)), and the mass of the heavy cluster is calculated using the law of mass conservation. The FFs charges are calculated according to the hypothesis of unchanged charge density. Configurations of the system after the first and the second ruptures are shown respectively in the fourth and the fifth columns of the table.
In the frame of the proposed scenario we have succeeded to reproduce the experimental energies of all three partners of ternary decay (Figure 2).
We assume that in contrast to conventional ternary fission the CCT occurs as a two-step decay of an extremely deformed prescission nuclear configuration in the valley of true ternary fission  or states associated with cold deformed fission in the binary channel . According to the neutron data, the population of such states reaches several percent.