 Hello everybody and welcome to the Latin American webinar of physics. I am Roberto Lineros. I'm going to be the host for this webinar. Today we have a very interesting talk. It's going to be about the data detection of dark matter. So first of all, remember that you can follow us in our social networks and especially in this YouTube channel. You are watching now the live transmission, but if you are watching it in the future, you can maybe consider to subscribe to the YouTube channel. So today's webinar is very interesting, as I said before, because it's going to be by Maria Lisa Sarza, she's a professor at the University of Zaragoza, in the center of the astro-particles and physics of the altosenergias from that university. She's going to talk about the Camp Frank underground laboratory and the title of her webinar is, Is the Dark Matter Wind Blowing at the Camp Frank Underground Laboratory? Welcome, Maria Lisa, and you can start this whenever you want. Thank you. So I'm going to share my slides. Okay, so you are seeing them. It's okay? Okay. Yes, okay. Perfect. So then it's my pleasure to be here and to present the recent results of the experiments I'm working in, that is the analysis experiment I will present you later. I want to thank the organizer for the opportunity to try to answer for you this question. Is the dark matter wind blowing at the Camp Frank underground laboratory? So to follow the talk more or less, I will cover the following topics. I will introduce the direct detection of dark matter and the annual modulation effect that is expected in the dark matter interaction rates. I will move them to the Dama Libra positive result, which has an associated long-standing puzzle covering really all the field of dark matter direct detection. I will present the Anais 112 experiment, that is the focus of my research, and the recently published result on the annual modulation corresponding to three years of data that have been very recently published in physical review there. This is the article that as you can see has been considered as editor suggestion and also featured in the physics magazine by the by the editors of the APA, the APS. So just a very brief introduction because probably all of you know about dark matter. The evidence on the existence of dark matter come from very different observational techniques, different scales, and different times of the universe evolution from galaxies to clusters of galaxies and larger scale structure, but also from the present back to the recombination and nuclear, the modern nuclear synthesis period of the universe. And we have a cosmological model that is very successful to explain all of this observation, but the expense of introducing a 27% of the energy and mass of the universe in the form of an invisible and non-baryonic matter, which is called called dark matter. This is the main topic, of course, how to look for this dark matter in the universe. There are a lot of candidates on a scene. They are covering orders of magnitude in mass and cross-section. We know very, very few things about this particle that we are searching for. But the most important thing is that none of the particles we know has the right properties. And so we have to look beyond the standard model of particle physics, what could be this dark matter. The best motivated candidates are coming from contexts different from the astrophysical context. So they in general are proposed for different parts of the particle physics. And I will focus on wings because they are a very convenient dark matter candidate. They are weak interacting massive particles, and they are really a very wide category of candidates. These particles have been thermally produced in the early universe, and they have the very nice property that frees out the annihilation cross-section that reminds the relic abundance that is compatible with the observed abundance of the dark matter. This is the wind-maya call, and it makes that for a GB particle in the scales of the weak interactions, we can explain the dark matter problem. Moreover, these candidates allow for direct detection and also indirect detection and also production at colliders because they are coupled to standard model particles. So you can see here in the diagram how these dark matter particles could annihilate and then produce cosmic rays, for instance, that could be searched for in satellites. But the dark matter particle could interact with the standard model particle and then produce a release of energy in our detectors. That is what I will call direct detection, but also two standard model particles interacting between them can produce dark matter particles at colliders. So this is a quite good thing for a candidate that could be detectable. And so this is the direct detection approach. Most of the experiments I will comment follow. For detecting dark matter, we need that the particle couple to the standard model particles by the somehow weak interaction and some energy could be released in the interaction with our detectors. Of course, we need very sensitive and radio-poor particle detectors because we expect this particle release a very, very small amount of energy and also that the interaction occurred very, very rarely in the detectors. Because of that, we have to shield our experiments against all the possible background contributions that could hide the signal we are searching for. And because of this, we are going to underground laboratories to seal the cosmic rays, but also introduce our detectors in very large shielding to protect against environmental radioactivity. And also, if possible, use the most innovative and successful background rejection techniques developing new technologies. But even in this case, how to be sure that we are finding the dark matter? How to be sure about a positive result of dark matter? For this, the key is to analyze a specific signature of the dark matter particle that is not shared by other backgrounds. And in this sense, I will commend the annual modulation, which is one of the most interesting features of these dark matter interactions. But there are also their possibilities. And it is very interesting, for instance, to develop detectors that are sensitive to the direction of the nuclear recoil produced by the interaction of the dark matter particle. Coming to this interaction of these whims with ordinary matter, as I have said before, a small transfer of energy in the form of a nuclear recoil is suspected in most of the models of the whims. But there are other options because we don't have really an idea of which is the interaction of the particle we are searching for. In any case, the interaction rate depends strongly on the specific wind particle properties. That means the mass and the interaction cross-section with the nuclei our detectors are made of. The dependence is also on the halo model, not only in the density of these particles in the galaxy, but also in the distribution of velocities at the position of the solar system that is where our detector is sitting. And this interaction rate is quite an involved combination of these dependencies particle and halo properties dependencies. And as I said before, which could be the interaction mechanism for the particles we are searching, the wind interaction rate will change strongly depending on this interaction cross-section, which is unknown and contains the details from the dark matter particle model at the microscopic level that I want to say that is dependent on the coupling of the dark matter particle with the standard model particle quarks and elementary particles. To arrive to this cross-section, we have to take into account the target nuclear structure and then to use the nuclear properties of the system in order to derive this total cross-section with our nuclei. In principle, the most general case can be described by means of an effective field theory having 15 possible operators that could describe the most general interaction between the particle that we have no idea what it is with the target nucleus. And this is quite complicated because these parameters have dependencies some of them in the velocity, but in some complicated way with the spin of the nucleus in very different shapes. The energy dependence, the energy rate is plot here, for instance, you can see here the standard spin independent term that is the O1 operator, which is the to consider and it produce energy release in our detector that has enormous exponential shape, not very easy to distinguish from backgrounds. For different mass of the particle we obtain different energy depositions, but other of these operators will produce very different spectral shapes for the energy deposition in the detectors because they depend on the velocity of the particles in quite an important way. That means that we don't have much an idea about what we are looking for in our detector. We could be the nuclear recoil energies we have to observe in our detectors. And this is more complicated when we try to compare experiments that are using different target nuclei because this rate has the dependencies quite introduced in the calculation. What is the usual procedure to compare experiment? We take a given interaction model for the WIMS. In general, we consider only this O1 operator that is the easiest one. We consider a spin independent interacting WIMS because the rate scales very easily with the square number of nucleons in the target. In this case, the comparison between experiments is quite easy because the scaling is easy. We have here the square reduced mass and the number of nucleons in the target for the comparison. And so we can plot the nucleon cross-section, but this is only one of the possible couplings between the Darmator particle and the nucleus we are studying. Even if it's only one of the possible operators, it's usually adopted to compare experiments because it's the easiest one. The same is happening with the Hallow model. How are the Darmator particles distributed in the Hallow? Well, we don't know. We have more or less some indication about the simulation of the in a particle and gravitational coupling in very large systems, but we don't have really a measurement of the Hallow distribution for the velocity of the Darmator particles. So because of that, to compare different experiments, we use the standard Hallow model, which assumes a spherical anisotropic distribution of the particles of the Darmator in the Hallow, and then a Maswellian distribution of velocities, which is truncated at the escape velocity of the Hallow. This is also a model-dependent assumption, and if we change this velocity distribution in the rates, we change the results. And we change in correlation with the target nucleus mass and properties in a quite complicated way. The Hallow is quite important because in general, experiments are only sensitive to the tail of the distribution above a given energy, a minimum energy or velocity of the Whims, because the threshold energy of the experiment, which is also different from one experiment to another, makes this dependence quite important in some cases. And you will hear later because this is important, but so yeah, I can now explain what is the annual modulation that we expect in the Darmator interaction rate. It's an effect because the Earth is moving around the Sun along the gear with orbital motion with a period of one gear, as you know. The Earth velocities are around 30 kilometers per second, but the ecliptic plane is tilted with respect to the galactic plane in which the Sun is moving around the center of our galaxy. And this is like we are moving inside the Darmator Hallow, and then we are moving through the Whims. So then is this because I put in the title of my talk the Whimped Wind. There is a wind of winds coming to the Earth to our detectors, and this wind is modulated along the gear. This wind is not blowing at the same rate in June and in December, because the velocity of the Earth with respect to the Sun is in the same direction or in the opposite direction along the gear, and then we add our velocity to that of the Sun or we are subtracting that. And this is not a large effect because we are seeing only about 13 kilometers per second more in June than in March, and 13 kilometers per second less in December. But this effect can be noticeable, and it produces a modulation in the interaction rate of particles. This is what we should observe for spinning the spin independent interacting winds in summer and in winter. It is a very, very slight difference, but if we compare, so if we look at the modulation amplitude that correspond to this difference between summer and winter, we will obtain a clear effect at different energies that is really very, very characteristic of the Darmator in the Hallow, because this period, the phase of the motion, and even the size of the effect is really not shared by backgrounds, and it's really a very distinctive feature of the Darmator interaction. As I told you, the effect is small. It should produce an inverse modulation at low energies. As you can see here, the modulation amplitude is expected to be positive for this spin independent interacting winds at the larger energies, but at a given point that depends on the mass of the nucleus and of the wind, the modulation should convert to negative modulation, and it's strongly dependent on the Hallow model. What we expect for the standard Hallow model is what we are showing here, but for another Hallow model, we should observe a different phase and a different magnitude of the effect. This is what an experiment in Italy, Damali Ibra, has been observing for more than 20 years. They have been observing a modulation in the interaction rate of serve by their detectors. Their detectors were at the beginning only 100 kilograms of sodium iodide detectors, starting the measurements in 1995, but they upgraded the experiment to 250 kilograms after 2003, and they are still taking data. The large results of Damali Ibra experiment were released in March 2018, and in this case, for the second phase of the experiment, they improved the performance of the experiment reducing the threshold to 1 KV, replacing all the PMTs, all the photomultipliers, with high quantum efficiency for the multipliers. The experiment has more than 20 years of data, and what they observe is really this kind of modulation in interaction rates that is here for only the 14 last years, so they have still more data before that, that have a very, very high statistically significant result for the presence of modulation. More than 12 sigma confidence level is what is attributed to this modulation signal of serve, and this modulation has the proper features that are expected for that matter, interaction modulation. They observe only the modulation at very low energies, below 6 KV of energy, not at the higher energies, as it should be produced. They don't observe the modulation for the backgrounds. For instance, when looking at events happening at the same time in two of the modules, the coincidences are attributed to background because the probability that the data matter particle produced two interactions at the same time in the detectors is really negligible. In principle, none of the systematics that have been proposed to explain this modulation is able to reproduce correctly the observation of the experiment of the malibra. This modulation, as I told you before, can be interpreted as produced by WIMPs, but this is a model dependent interpretation. In this case, I have chosen a model and in this case, for instance, the spin independent is a spin-conserving interacting WIMPs and a standard HALO. In order to find which parameters of the WIMP, in terms of mass of the particle on coupling to the nucleons, are compatible with the darn matter libre result. We obtain here these two regions in the parameter space, depending on the rate is dominated by sodium interaction or by iodine interactions. This is the compatibility with the malibra result and which is the experimental situation. The experimental situation is that only one experiment, the malibra, observes a model independent annual modulation compatible with darn matter and moreover in the standard HALO. But there are other much sensitive experiments that don't observe any hint of the interaction of these darn matter particles and the comparison is more than dependent. For your information, the most recent comparison done in the frame of the APEC committee report that was released last April, very recently, you can see here the darn matter libre regions I showed you before, and the results of many other very sensitive experiments around the world that don't observe anything compatible with the malibra. These lines show the splash on plots, the region of the parameters that are excluded by these experiments that are not compatible with the darn matter libre result. So this makes a strong tension not only for these candidates but also assuming more general and HALO interaction models. However, as far as we don't know the properties of the particles we are searching for, this comparison is always more than dependent. And this is a problem because we require a model independent confirmation or refutation of the malibra result. Is mandatory in order to you to answer the question, is the malibra result a door into new physics or is just systematic. To answer this question using the same target that the malibra has used is really one of the main points for doing more direct comparison. But it's not the only thing. Of course, using the same target, so you mayodide, as I said, is really very important because we could reduce many dependencies on the coupling to the nucleus of the WIMP, although these uncertainties are almost released. But a good knowledge of the detector response is required. And I will comment later on this question a little more. You can see here, according to the committee report, which is the status of the experiments of sodium aiodide that are presently running in preparation. You can see here, of course, the malibra is still taking data and running, there are two experiments, Anais 112 and Cosine, and there are some experiments in preparation. Just to say a few words about Anais and Cosine that they are both in data-taking phase, Anais has 112 kilograms is less than the malibra. Cosine is only using 60 kilograms of mass effective mass. They have a larger mass, but this is only the effective mass for the annual modulation search. They are taking data since 2016, and we are taking data since 2017. And then we are trying to confirm or refute the malibra result in the next few years. But other projects in preparation are, for instance, Sabre experiment, which is developing more radio-poor crystals, and they plan to use two detector sites, one of the north hemisphere in the Gran Sasso laboratory, another in Australia in the Stauel laboratory that is in construction, and also the Cosinus experiment that is also very interesting because it uses a different technique. They are planning to use volumeters of sodium aiodide in order to search better for this different response to nucleary molecules that is produced or that is expected in scintillation detectors. So concerning Anais, Anais stands for annual modulation with sodium aiodide scintillators and the goal is the confirmation of the malibra result with the same target and technique. But of course our experimental approach is different and this will be the focus of the rest of my talk. And the environmental conditions that would affect systematics are also different. We are working at the Canfrán underground laboratory that is in Spain under 2450 meters of water equivalent and we are taking data since August 2017. We use a three times three matrix of modules, 12.5 kilogram each that are cylindrical in shape and made by Alpha Spectra Company in the States. We have coupled in a second state but Canfrán clean room, high quantum efficiency photomultipliers from Hamamatsu and our data analysis that is also a very important part of our work is blind. The region of interest was blind from beginning so I will comment a little more. Concerning Canfrán is a laboratory under the Spanish Pyrenees in a tunnel, the Somportana, which is connecting Spain to France with a rock over a garden of 150 meters. That is a smaller rock over a garden than Gran Sasso laboratory. You can see here the death of Gran Sasso and this implies that for instance all the possible systematics related with my own residual flux could be different and should affect in a different way our experiment than that of the Malibra. The relevant experimental features of Anais, so for instance we work in collaboration with the company producing the crystals to build in mylar windows in order to allow for low energy calibration. To have a robust energy calibration at the very low energies where the signal is expected is one of the most important points in the analysis of our experiment. We can calibrate with cadmium 109 sources mounted on flexible wires that are introduced and taken out our shielding periodically. Every two weeks we can calibrate simultaneously all the nine modules in our experimental setup. Our crystals have sewn an excellent light collection. Cintillation is produced in the crystal by the energy deposition of a particle and then we have to collect this library of multipliers with the highest possible efficiency. In our modules the light is collected at the level of 15 photoelectrons per kb with a very very large and homogeneous distribution of the light collection between all the modules. The lowest one has 12.7 photoelectrons per kb and the highest one has 16 photoelectrons per kb. If you compare with the Malibra results even in the second phase which is the improved phase of the experiment their light collection is between 6 and 10 photoelectrons per kb. This implies our detectors collected much better the light produced by the particles and then our threshold energy and our response should be more sound should be better than that of the Malibra. We are monitoring continuously and on the data taking this light collection which is changing along the time but we can correct by these drifts that we are still trying to understand by calibration and correctly correcting any gain drift in the detectors. However the total light collection is quite stable in all of our modules after the three years and all of these parameters and the evolution of the parameters has been deeply studied in our last results paper. I commented you before. This is our setup but these are some pictures of the building. You can see here from outside to inside we have polyethylene and water to moderate neutrons and active muon beto consisting of 16 plastic ventilators, a tight box preventing the radon entrance into the shielding, 20 centimeters of left with low activity and the most internal part of our shielding are 2 centimeters of archaeological left. All of these is our shielding to prevent environmental radioactivity, produce interference with our detection of the of the matter that is our goal and in principle this is different from the Malibra that has no active beto against the residual muon flags. Of course in comparison it's more important because as I told you before our residual muon flags is higher than in Gran Sasso. This was the commissioning in March 2017 and the data taking started in August 2017. Our data acquisition system has been very well tested with previous prototypes. We have the two signals from every pre-empty after a particle interacting in the crystal is produced in a light that is reaching the photomultipliers and then the signal is processed by very nice matak digitizers with a high resolution 14 bits and two gigasamples per second and we trigger with the an between the two photomultiplier signals in a 200 nanosecond window. Our system is really very free of noise and is not affected by vibration of electrical contribution but as I will comment you our problem is different one of the threshold position. So with these signals produced by by our modules in this way by this data acquisition system we have to calibrate very well the light and to convert it into energy and we do the calibration at the region of interest with a very high accuracy combining the periodic external calibration lines of cadmium. This is the cadmium spectra obtaining the nine modules you can observe three lines below 100 keV at 18 22 and 12 kilo electron volts and this is done every two weeks and we combine this external calibration with potassium 40 and sodium 22 internal calibration contamination background lines at 3.2 and 0.9 kilo electron volts as background lines have a slower have a smaller statistics we use a very 1.5 months of accumulation these these events from the background in order to calibrate our our experiment and so then the region of interest is calibrated with four lines below 25 keV in order to take into account the non-linear behavior of sodium iodide below 100 keV and so our calibration is quite good in this regime and these two lines are fully in the region of interest of our experiment so then we can say that we are calibrating very very nicely our experiment these are the lines accumulating the time of measurement 3.2 and 0.9 kilo electron volts selected by the coincidence with high energy gamma in a second module and so then we are not affected by any problem with the triggering because it's done by the high energy background gamma line and we demonstrate that we are triggering below 1 keV you can see here not only the line of sodium 22 but also the L capture in sodium in potassium 40 decay we are observing events below 1 keV very nicely but this is not our threshold energy because of other problems with the with the experiment analysis that I will focus later so the calibration is determined with these cardinal lines in the background and we can correct any gain drift with this with this calibration procedure in a very robust way and so after calibration we can develop our analysis strategy we have done this with the first year of data we blind the events of interest that are even is the 1 to 6 keV energy region and only single heat events and the coincidence events between modules and for determining the fine tuning analysis parameters and efficiencies for the selection procedures we are using only m2 events in the ROE which are youtube background for instance this potassium 40 and sodium 22 events but also the cadmium calibration events and we are blind 10% of the first year of data for background assessment but all this procedure was published before doing the first analysis in this 2019 paper in the european physics journal c with the first year analysis we fixed the selection criteria and we have kept this criteria for subsequent analysis we have just updated efficiencies but not changed the parameters of the filtering and this is our filtering procedure so our rate of events in the region of interest is dominated by non-sintillation of sodium iodide events so we have to apply a stronger event selection procedure first we select only single heat events to reject the backgrounds produced by radioactivity in the environment we select also events arriving more than one second after the moon interacting in the beta system this is because we want to reject all the possible moon related events in our detectors and then we have to remove this non-compatible with blue scintillation events they are some events that are too fast to be produced by scintillation in sodium iodide that has a scintillation constant of 250 ns on second but also two slow events that are associated with tails of pulses and you use a v parametric cut considering the tail of the scintillation pulse with respect to the total scintillation pulse and the distribution of the mean photoelectron arrived by times moreover we have to reject asymmetric events that means events that have a strong signal in one photo multiplier but a weak signal in the second photo multiplier these events are not compatible with the bull scintillation in sodium iodide and we established the cut at the number of photoelectrons the peaks for instance here you can see two peaks we require more than four photoelectrons in every PMT of every module in order to keep the event as produced by as good scintillation events so you can see here this is our total rate draw data after applying this same behavior and v parametric cut we reduce very strongly the events that are passing the cut and you can observe here the peak that is appearing in our data this is the potassium 40 background line that won't be observed if we should keep only raw data in the region of interest so this is all the cuts we are applying and you can observe really the effect below 10 kb that is really the reducing strongly the interactions that we observe in our detectors and these criteria of selection of events are the motivation of the analysis threshold we have to establish but the parameters for the filtering where established as told you before before unbinding with the first year of data and using only population of events outside the region of interest these are the efficiencies we have calculated with the cadmium sodium and potassium events that fix our analysis threshold at one kb because we require that our efficiencies for keeping good scintillation events are larger than 10 percent and for that we have to go only down to one kb the efficiencies are calculated detector by detector are larger than 90 about 2.5 kb below two kb they strongly reduce down to one kb which are at the level of 15 and 25 percent depending on the detector module so we use these efficiencies to correct of course our even rate after filtering in order to recover the estimated even rate at every energy but we can check the the goodness of this efficiency and the good the energy of calibration stability for instance using potassium and sodium population that are very well tagged by the high energy gamma that is triggering the detection in a second module and then we are almost free from a noise in this population you can see here the events compatible with potassium 40 the evolution of the rate in the three years of data and this is for sodium 22 behavior along the three years in potassium we observe a constant rate compatible with the half life of the isotope because that is of the order of billion of years and in the case of sodium 22 we obtain a decay an exponential decay that is compatible with the half life of the sodium 22 isotope so this is really a very good indication that our checking of good events is quite well established at the threshold because these events are in the one kb energy region so before presenting results I have to comment that despite the large effort of analysis team looking for ultra-poor sodium iodine detectors our background is worse than the malibra our background in the region of interest is dominated by the crystal radioactivity we have higher potassium for the level and higher lead to 10 level in our crystals than the malibra has this is the main problem in our region of interest but I have to stress that the quality of the macrystals has has been still not reached by any experiment sabrian cosine collaboration are working strongly and progressing towards more radio poor crystal growing but they are still having troubles with lead lead to 10 for instance so this is our background is worse than the malibra we are higher value than two counts per kb kilogram and a between two and four in the region of interest but we have a very good background model so we understand our background is not good but it's understood in all the energy regions it can be explained but natural change led to 10 potassium 40 and 3dium but we underestimate the right in the one to two kb energy region you can see here more clearly in this plot and we are working to understand this underestimate of this region of interest below two kb but although this is this is quite important because it affects our region of interest and we are working on matching learning techniques trying to reduce the possible leaking in our filtering protocols because we think that this stress of events is related with our selection criteria so we are trying to improve it we remark that constant backgrounds should don't affect the annual modulation analysis and we can say this because our model predicts very well that the evolution of the background detector by detector you can see here which are the evolution of the rates that we expect in three years so no here is for more than three years three years or more or less here we can predict the behavior of our rates and we are producing satisfactorily the time evolution of cell for instance for background events in the one to six kb region this is just adjusting to a factor f our background model that is just at a few percent of the unity and here is for events in the outside the region of interest between six and ten kb that is also in the few percent level of our estimates of our background model so we can use these to present our results on our final modulation we published our first results in 2019 in physical review letters but now we have updated for more time for more exposure 300 kilograms per year with a very excellent duty cycle 95 percent of our time is lifetime and most of the downtime of our experiment is due to the calibration that are done every two weeks so this is a very very nice performance of our experiment and we are using for the present results more than 1000 days of data after removing me on tag related events with respect to the previous analysis we have improved our background modeling and we have checked also some systematics and the consistency of the result and we have simulated to see the experiments by Monte Carlo to analyze possible bias and checking the sensitivity in our experiment so what we do is to minimize the number of events that we observe in 10 days 10 days being enough our data considering a poison disturb uncertainty of course and correcting conveniently by the lifetime of every beam and the efficiency for every energy region in the case of this expected number of events miss me we are considering a background modeling that i will comment now and a term due to the modulation expected for the for the tar matter signal that has an amplitude that should be fixed to zero in the new modulation hypothesis and should be free in the modulation hypothesis study the frequency will be fixed to one year period and the phase should be fixed also to the second of june for the maximum of the modulation as corresponding to the standard hollow model in order to compare with the malibra experiment so for the background modeling we have considered two different background modelings combining the detectors of the nine modules together and exponentially decaying background we have three parameters our zero is the background parameters that is including also the possible signal of the matter that is averaging time tau and f take into account the contribution for the for the background that is exponentially decaying but here could be also any a constant in time background contributing as i commented before in the region below to kb a constant background should be included also in this rco parameter thanks to this term in the background modeling function we have also taken the probability distribution function from the background modeling that i showed you before corrected by a factor f and the r0 parameter which is also led free as noise parameters in our fitting we obtained these results for the two modeling of the background and for the new hypothesis and the modulation hypothesis you can see here the results for the p values and for the modulation amplitudes that we are fitting i will show you better later these values we obtained better p values in the two to six kb region and in the one to six kb region i will comment a little in this later but this is the comparison with the values obtained for the modulation of the malibra so here we have fixed the modulation amplitude to the malibra result and you can see here how the p value is much worse in both scenarios in both energy region with respect to the new hypothesis and to our best fit best fit to a modulation hypothesis the third modeling considers every detector individually and can take into account possible symptomatic effects related with the different backgrounds and efficiency of the different modules that we can observe here in this case the results are more difficult to see but we reproduce quite nicely the behavior of all the detectors some detectors have very low p values we think that these detectors are the problem of the global low p values in the one to six kb energy region and we think that these detectors are affected by a worse filtering procedures selection criteria and we are trying to correct this in the future for other analysis however this analysis combined for all the crystals has the slide the slower the smallest sorry standard deviation and this is the value fully compatible with all the other values derived from the other techniques that we have we have taken for comparing and presenting our results with respect to the malibra this is the modulation derived from our fitting in black points derived with respect to the malibra result that is shown in blue you can observe how our best fits are incompatible with the malibra result at 3.3 sigma level and 2.6 sigma level in the two energy region that we are considering for a sensitivity at the 2.5 and 2.7 level in this two energy region the sensitivity is plot as the the green yellow and blue lines in this in this figure and this is what we expected for our a priori sensitivity estimates so these were our estimates in terms of sigma confidence level with respect to the malibra and we should be at 3 sigma from the malibra result within the two scheduled five years of data taking so we think that we are really very nicely reproducing our estimates and that we can be at this level next year along next year the time we could need for the analysis of the of the results concerning our checking just a brief comment we have checked the VIN size changing from 5 to 30 days negligible effect on the results for the modulation we have a check possible bias with our Monte Carlo so the experiments modulation we observe no bias and our result is fully compatible with the standard deviation derived from the Monte Carlos we have compared one two years with respect to two three years and results are fully compatible we have also released the phase of the modulation and then we have observed with respect to dama that our result is bias but any modulation search in this parameter space should be bias and if we take into account this bias we observe our result compatible at one sigma with zero in all the with the absence of modulation in all the phase of the parameter space and we have also released the frequency no statistically significant modulation at any frequency is observed in our data so in principle all of our results and tests are compatible with the absence of modulation so I conclude by saying that we are searching for an one modulation since august 2017 with 112 kilogram of sodium iodide come from and our experiment is ruining quite smoothly with very nice experimental features with three years of data the new hypothesis is well supported by all of our analysis and the best fits are incompatible with the malibra for the sensitivity more than 2.5 by less than three sigma now but the best fits are incompatible already at three sigma level in the 2126 kV energy region so we confirm our sensitivity projections and we expect to reach in 2022 three sigma level of the malibra test however before ending and let me now a few words about is this really a modern independent testing of the malibra result and I hope you make some question about this issue because using same target material should be just enough the comparison is more direct but the response of both detectors has to be very very well known and we don't know so well the response function of sodium iodide especially for nuclear recoils this is because of the quenching factor the scintillation produced by nuclear recoils is quenched with respect to electron recoils and this factor is required for the calibration scale to convert energy in terms of electron equivalent energies to nuclear recoil energies and this is the present situation for sodium iodide quenching factors this is the dispersion of experimental results that are present now in the literature so there are still too many uncertainties in the quenching factors for sodium iodide and this would be improved in order to say that two experiments as the malibra and anis are the compatible or incompatible in our case we are measuring our quenching factor for different crystals and we are trying to determine this quenching factor at the best level we expect to have results very soon but the comparison with the malibra will be still dependent on the results for the malibra crystals we are also studying possible explication of the modulation signal in terms of for instance the muon possible modulation that has been suggested to explain the malibra many times but the drama reply is robust the phase of the modulation is inconsistent and the muon interactions don't fulfill the darmator requisites however we are observing quite a lot of interaction related with muons that we are removing but we are trying to understand better how these events could be contributing in the low energies and so we expect also to do some works in this direction in the next future so how can we do a test of a damalibra experiment with less time and less mass of detection and also with a worse background I have to say just that the modulation amplitude is large but our sensitivity should be enough to distance between three and four sigma confidence level to go beyond lowering the background is a must and then we have to develop a lower background crystals higher radiopurity crystals of sodium iodide and the most important things we plan to do our data public to allow for independent analysis because we really are convinced that sharing the data and procedures is the way to find a solution to this puzzle thanks a lot for your attention thank you very much Marisa for the for the nice talk and for the nice results from Anais experiment so for the for the people that is following the the live transmission you can write your questions in the youtube chat for the moment we are going to start with maybe some question from here for the audience in the zoom session so if somebody has a question for Marisa we can start with with that question so I okay I'm going to start with Marisa when you mentioned the okay that Anais can contrast the results of the damalibra three sigma it's just now that you have this three years of period but at most you expect that Anais can rule out at four sigmas that is a matter of statistics or yeah it's a matter of statistics with time we improve but so in our estimates to reach five sigma we need 10 years and then this is maybe too much so we are not really deciding yet how to finish our measurement but it will depend if there are a better experiment running or not yet at the moment I think that we are the with cosine we are the only experiments that are taking data and then okay we can take data for some more years but really to get to five sigma uh it's a very very long time so we are speaking about 2020-17 yeah so another question with regarding in the case of the damalibra modulation is just a systematic issue I mean it's not that matter it's something else is there is a way to to contrast these points that we other I don't know because I know that point that kanfrank is it's not in the same latitude that the gran Sasso but but almost like if it is atmospheric yeah so in so in our opinion if it's atmospheric but strange it should be similar to to gran Sasso if it's atmospheric but related with mions then it should be much worse in kanfrank because we have a low rock over burden that is lower and so then if it's related for instance with a neon modulation then we should see the effect it's true that we are removing these events in our data now but we are not observing actually our effect in the modulation of mions in in our case so then we are working in this direction because it's one of the systematics that is more easy to to face on our point of view other possible systematics are really much more complicated to to contrast because of course our experimental approaches are the same and and so then there are possibilities in the in the measurements that that are not taken into account by by the other experiment in the same way so I will say that there is place to study systematics but for that the best way will be that the Malibra will share their data and the other people look at their data in different ways and that more information of their acquisition and and protocols for filtering for instance the filtering is very important with all the experiments working in these energy regions and the filtering there is not much information on how it's done or which even are removed for kept by by the Malibra so but but it's not easy from our side maybe to find the systematic that is explaining the result okay thanks I don't know if there is another question from here for the for the audience otherwise we can go to some question from youtube because we have a couple first we have one from from Karen MC that she's asking how useful is animal modulation to differentiate between a wind signal and the neutrino background and higher energy like atmospheric or the dsnb benefit no of course a annual modulation will be very useful if a signal is found so I will say that the the most important question so it's probably at this at this rate of improvement in backgrounds in the experiments of that matter probably neutrinos will be detected before that matter and so if there is enough statistically speaking events so of course they should be searched for all the possibilities so I mean for neutrinos there will be also some modulation for the solar neutrinos because the distance to the sun is not the same along the gear although the difference is is quite small but I will say that for that the most important problem in this case to compare these effects will be the statistics always but it will be a very distinctive feature because the amplitude of the modulation is dependent of course on the on the relative velocity with respect to the halo and this is not shared by atmospheric or solar neutrinos backgrounds but but it's involved of course there are many modulations in when these neutrinos are detected but the directionality will be also quite very important think probably this is the most interesting experimental issue to be sensitive to direction not not in global to the annual modulation okay so there is a it's not a question but it's a comment from paolo salucci he's saying that well first of all he's congratulating to you the same very very interesting result not crucial but it is fair to not to claim a standard halo model with that work on it have got an agreement even at the level of the gas core issue problem so he's more or less commenting I mean trying to about yeah when when they compare all the direct detection experiment with the same halo model of course if it is maybe it could be a little some bias in that of course now that is clear that this implies an important bias and for some and in the case of the annual modulation even more than the comparison of the of the normal cross-section of interaction the normal interaction rate and but so it's very difficult to make comparisons halo independent and so there are some some proposals and some ways to try to do comparisons so in our case of course the question is that to test the maribra modulation we have to assume the same the maribra modulation result is compatible with the standard halo modulation yeah but but but it's true that if we don't know how is the halo we should face the problem on the contrary if we detect an annual modulation we could use the result to test the halo models yeah so and also in some way yeah also in that sense since the damalibra is claiming another modulation they are claiming with the standard of course halo yeah so yeah at least not halo is compatible with the maribra modulation exactly so there is no no issue related to the other motivation itself i mean you are comparing the same halo model with the same kind of initial setup so this is because of that our comparison is more than independent because so we put the same detector with the same target in the same place we should observe the same modulation so we are not assuming anything about interaction mechanism or the halo model the only problem that i mean with our comparison with the maribra is really the energy scale if the point in factor for instance is not the same for the maribra detectors and for our detectors our energy region from one to six tb could be different because we would have there an uncertainty in the scaling of the detector but this is a quite very different problem with respect to the comparison with other experiments where the model dependencies are really inside of the calculation of the interaction rates and so because of that we think that our comparison is important and we should check really these these very small questions that are still to be fixed yes so i don't know if there are other questions because also we have some another question about the the how is the damalibra damalibra sorry and it's for the for the for the total cross-section regarding the it's been it's been independent cross-section like our analysis yes so i mean not the animal modulation but if it is planned for the so in principle our background is really much worse than the background of most of the experiments in the in the field so so i have said before i will show you again the plot of the comparison between the different experiments searching for the armata so we are not competitive at all with the most of the armata searches because these detectors have a have a background which is quite large also damalibra background is large with respect to most of the most sensitive experiments searching for the armata so we are not aiming at doing a real exclation in in the parameter space because we cannot compete with most of the experiments in the field but but in principle our sensitivity should be similar to that of cosine experiment you can see here the plot because our background is really very similar to that of cosine so in principle we should be at this level at the level of the amalibra signal that is at the level of our backgrounds that that in all the sodium iodide present experiment is at this level but so in principle we have not tried up to now to to do this more independent effort because we think that is not really interesting to compare with amalibra it's much more interesting focus on our annual modulation result which is more than independent this plot is more than dependent also for sodium iodide because we are fixing an interaction cross-section scheme and and then but but we are not better than 10 to the minus 42 centimetre square of cross-section so I mean this is really limited by the background and and so then this is our this is our limitation we cannot go far that if we don't change the crystals I mean yeah also if you don't increase the volume of okay yeah so and a small question also because many of these other experiments they try to use also the cross-section regarding two electrons in in the case of crystal can be done this type of analysis so it's because usually it's for phenol kind of liquid they are using so they have considered this migral effect I probably they refer to this question know that that they use the energy released to the electrons in the in the material in order to in order to to get also a information at the lower energies but so in principle in our case we will we have focused up to now so of course this will take it this into account for the final analysis with the five years in order to take all the possible take out all the information possible from our data but but I mean for the comparison with amalibra it shouldn't be important because if this effect is happening it's happening in the amalibra detectors also and also in our detectors so we are looking globally at the scintillation light produced by any particle that is releasing energy in the detectors so so in principle for our comparison I wish I should say that this is not much important it's important to do this analysis in the in the model-dependent way and well okay we will try to look if this we could be sensitive to some of these proposed effects that have not been confirmed by measurements that have been only proposed to be there but I should say that it's not the the most important thing in our case that we are so background limited that we are really very very low sensitive to to this kind of of a special effects sent on experiments are are really in other order of magnitude of a ground so you can see here the several orders of magnitude they can reach in cross-section okay so I don't know in youtube we don't have other questions I mean paolo salucci is saying thanks for the when you answered I don't know if there are other questions from here for the audience if they want to ask to marisa otherwise we are in the also we are in the time to to close the transmission so we we thanks marisa for such a nice webinar and such an important result with the analysis experiment and for the for the people that are following us in in youtube please you can consider to subscribe also to be updated I mean to keep up update with the with the following webinar that is more or less in in two weeks the webinar 117 and for the moment we can say goodbye to everybody and see you in the next time in the physics yeah